Rella

rella

Blogging has become a powerful tool for business and personal brand building. With the rise of digital and social media platforms, bloggers have become influencers in their own right, with many earning a substantial income from their online presence. In this article, we highlight three main affiliate marketing strategies from five highly successful influencers and break down how you can benefit from these tactics, too. 1. Leverage CPA and Commission Opportunities Sami Clarke is a wellness blogger with 471,000 YouTube followers. She discusses a healthy lifestyle, workout routines, recipes, and nutritional choices. Her digital platform, FORM, provides a subscription to workouts, nutrition plans, and other wellness content with the goal of "encouraging you to be the best version of yourself." Affiliate marketing through cost per action (CPA) is one way Sami monetizes her blog. This involves promoting products or services on her blog and receiving a commission for every sale made through her affiliate link. Bloggers can also earn CPA income by generating leads through newsletter signups or other actions a brand is interested in. With a large loyal following on both YouTube and Instagram, she first established trust by delivering free content before promoting any products or services. This is an important tip to remember, especially for new influencers. Becoming a brand partner or ambassador is another common way influencers generate revenue. For example, Sami also partners with wellness and fitness brands like Gymshark and Alo Yoga, adding affiliate links in her posts and encouraging followers to use them when they make a purchase. In return, she earns a commission on the sale. 2. Offer Value Beyond Basic Affiliate Marketing Codes Beauty and fashion blogger Christine Lee has built a following of 256,000 viewers on YouTube thanks to her fresh, exciting content. Not only does she use commissionable links, but she also offers valuable free tutorials (like her popular video "How to pose for Instagram photos") and fun giveaways to incentivize her followers to make purchases. Additionally, Christine creates specially designed Lightroom presets that followers can purchase to achieve a specific look or style in their own photos. Natalie Barbou, lifestyle influencer and founder of Rella, an all-in-one management tool for influencers, has also created her own platform called Let's Collaborate, which connects brands with bloggers in their niche and provides bloggers with easy tracking tools to manage collabs and earnings. A successful entrepreneur, she also monetizes her blog by partnering with brands like H&M and Benefit Cosmetics for affiliate commissions and making aesthetic posts and stories that showcase various looks, allowing viewers to go straight to her blog to purchase their favorite looks. New York-based fashion stylist Liz Teich (@TheNewYorkStylist) takes a similar approach to make money with her Instagram account. She offers numerous services like closet refreshers and generous discounts with partners, such as 40% off Rent the Runway subscriptions. However, the distinctive style and shoppability of her posts set her apart from others in her niche. She creates a range of content, including outfit photos, fashion and styling advice, and behind-the-scenes glimpses into her life as a stylist and blogger. Her posts are well-curated and visually appealing, focusing on high-quality photography and engaging captions. As a stylist, she works with fashion magazines and puts effort into strengthening her personal brand as well, garnering features in Forbes, the New York Post, and other prominent publications. She also has a link page that makes it extremely easy to create a sales funnel from her Instagram to a brand website to buy any look or product she shows on Instagram, streamlining the process of generating revenue. 3. Take Advantage of Direct Advertising The "queen of affiliate marketing," Miss Jemima is a beauty blogger with over 183,000 YouTube followers. Her primary strategy is actively using direct advertising and affiliate links to monetize her content. The effectiveness of her plan lies in her understanding of how to get her community engaged with her content. Direct sales is the most useful tactic for people who are great at creating connected communities who are eager to try products endorsed by their favorite influencers. For instance, Miss Jemima most often uses direct advertising when reviewing tools or products and includes an affiliate link to the product in the description. If viewers click the link and make a purchase, she earns a commission on the sale. She also uses all-caps captions and clickbait on videos to gain more attention, boosting the chances of creating viral content. Her approach is also a great example of using gamification like cooking challenges to encourage likes, shares, and involvement. Christine Lee also makes the most of her direct advertising partnerships. In a recent example, she published a makeup tutorial using Tarte products and included affiliate links to the products in the video description. The campaign helped drive sales of Tarte's products and raise brand awareness among her followers. Remember, creating value for brand partners is just as important as giving followers extra value. Unlock Your Full Earning Potential By Leveling Up Your Affiliate Marketing Strategies Successful bloggers have found ways to generate income from their online presence by using a range of sales tools, such as CPA, affiliate links, promo codes, and direct advertising from brands. They also understand the importance of establishing trust within their niche, developing an appealing personal brand, and providing value to both partners and followers. Analyzing the techniques employed by these bloggers can help you learn how to leverage sales tools effectively in your own blog and grow your online business. Whether you're a novice or an experienced blogger, these top influencers offer inspiration and a wealth of knowledge to aid your journey. The benefits of affiliate marketing for bloggers are evident, and with the right approach, you can also achieve success online.

Learn How to Monetize Your Blog From These 5 Lifestyle Bloggers

Types Of Medical Databases And Their Benefits For Humanity

Assessing VASP Solvency For Cryptoassets: Background and related literature

Authors:
(1) Michael Moor, Department of Computer Science, Stanford University, Stanford, USA and these authors contributed equally to this work;
(2) Qian Huang, Department of Computer Science, Stanford University, Stanford, USA and these authors contributed equally to this work;
(3) Shirley Wu, Department of Computer Science, Stanford University, Stanford, USA;
(4) Michihiro Yasunaga, Department of Computer Science, Stanford University, Stanford, USA;
(5) Cyril Zakka, Department of Cardiothoracic Surgery, Stanford Medicine, Stanford, USA;
(6) Yash Dalmia, Department of Computer Science, Stanford University, Stanford, USA;
(7) Eduardo Pontes Reis, Hospital Israelita Albert Einstein, Sao Paulo, Brazil;
(8) Pranav Rajpurkar, Department of Biomedical Informatics, Harvard Medical School, Boston, USA;
(9) Jure Leskovec, Department of Computer Science, Stanford University, Stanford, USA. Table of Links Abstract and 1 Introduction 2 Related Works 3 Med-Flamingo 4 Evaluation 5 Results 6 Discussion, Acknowledgments, and References A Appendix 2 RELATED WORKS The success of large language models (LLMs) Brown et al.; Liang et al. (2022); Qin et al. (2023) has led to significant advancements in training specialized models for the medical domain. This has resulted in the emergence of various models, including BioBERT Lee et al. (2020), ClinicalBERT Huang et al. (2019), PubMedBERT Gu et al. (2021), BioLinkBERT Yasunaga et al. (b), DRAGON Yasunaga et al. (a), BioMedLM Bolton et al., BioGPT Luo et al. (2022), and Med-PaLM Singhal et al.. Although these medical language models are typically smaller than general-purpose LLMs like GPT-3 Brown et al., they can match or even surpass their performance on medical tasks, such as medical question answering. Recently, there has been a growing interest in extending language models to handle vision-language multimodal data and tasks Su et al. (2019); Ramesh et al.; Alayrac et al. (2022); Aghajanyan et al.; Yasunaga et al. (2023). Furthermore, many medical applications involve multimodal information, such as radiology tasks that require the analysis of both X-ray images and radiology reports Tiu et al. (2022). Motivated by these factors, we present a medical vision-language model (VLM). Existing medical VLMs include BiomedCLIP Zhang et al. (2023a), MedVINT Zhang et al. (2023b). While BiomedCLIP is an encoder-only model, our focus lies in developing a generative VLM, demonstrating superior performance compared to MedVINT. Finally, Llava-Med is another recent medical generative VLM Li et al. (2023), however the model was not yet available for benchmarking. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Michael Moor, Department of Computer Science, Stanford University, Stanford, USA and these authors contributed equally to this work; (2) Qian Huang, Department of Computer Science, Stanford University, Stanford, USA and these authors contributed equally to this work; (3) Shirley Wu, Department of Computer Science, Stanford University, Stanford, USA; (4) Michihiro Yasunaga, Department of Computer Science, Stanford University, Stanford, USA; (5) Cyril Zakka, Department of Cardiothoracic Surgery, Stanford Medicine, Stanford, USA; (6) Yash Dalmia, Department of Computer Science, Stanford University, Stanford, USA; (7) Eduardo Pontes Reis, Hospital Israelita Albert Einstein, Sao Paulo, Brazil; (8) Pranav Rajpurkar, Department of Biomedical Informatics, Harvard Medical School, Boston, USA; (9) Jure Leskovec, Department of Computer Science, Stanford University, Stanford, USA. Authors: Authors: (1) Michael Moor, Department of Computer Science, Stanford University, Stanford, USA and these authors contributed equally to this work; (2) Qian Huang, Department of Computer Science, Stanford University, Stanford, USA and these authors contributed equally to this work; (3) Shirley Wu, Department of Computer Science, Stanford University, Stanford, USA; (4) Michihiro Yasunaga, Department of Computer Science, Stanford University, Stanford, USA; (5) Cyril Zakka, Department of Cardiothoracic Surgery, Stanford Medicine, Stanford, USA; (6) Yash Dalmia, Department of Computer Science, Stanford University, Stanford, USA; (7) Eduardo Pontes Reis, Hospital Israelita Albert Einstein, Sao Paulo, Brazil; (8) Pranav Rajpurkar, Department of Biomedical Informatics, Harvard Medical School, Boston, USA; (9) Jure Leskovec, Department of Computer Science, Stanford University, Stanford, USA. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 Related Works 2 Related Works 3 Med-Flamingo 3 Med-Flamingo 4 Evaluation 4 Evaluation 5 Results 5 Results 6 Discussion, Acknowledgments, and References 6 Discussion, Acknowledgments, and References A Appendix A Appendix 2 RELATED WORKS The success of large language models (LLMs) Brown et al.; Liang et al. (2022); Qin et al. (2023) has led to significant advancements in training specialized models for the medical domain. This has resulted in the emergence of various models, including BioBERT Lee et al. (2020), ClinicalBERT Huang et al. (2019), PubMedBERT Gu et al. (2021), BioLinkBERT Yasunaga et al. (b), DRAGON Yasunaga et al. (a), BioMedLM Bolton et al., BioGPT Luo et al. (2022), and Med-PaLM Singhal et al.. Although these medical language models are typically smaller than general-purpose LLMs like GPT-3 Brown et al., they can match or even surpass their performance on medical tasks, such as medical question answering. Recently, there has been a growing interest in extending language models to handle vision-language multimodal data and tasks Su et al. (2019); Ramesh et al.; Alayrac et al. (2022); Aghajanyan et al.; Yasunaga et al. (2023). Furthermore, many medical applications involve multimodal information, such as radiology tasks that require the analysis of both X-ray images and radiology reports Tiu et al. (2022). Motivated by these factors, we present a medical vision-language model (VLM). Existing medical VLMs include BiomedCLIP Zhang et al. (2023a), MedVINT Zhang et al. (2023b). While BiomedCLIP is an encoder-only model, our focus lies in developing a generative VLM, demonstrating superior performance compared to MedVINT. Finally, Llava-Med is another recent medical generative VLM Li et al. (2023), however the model was not yet available for benchmarking. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. available on arxiv

Med-Flamingo: a Multimodal Medical Few-shot Learner - Related Works

Authors:
(1) Bobby He, Department of Computer Science, ETH Zurich (Correspondence to: bobby.he@inf.ethz.ch.);
(2) Thomas Hofmann, Department of Computer Science, ETH Zurich. Table of Links Abstract and Introduction Related Work Preliminaries Simplifying Transformer Blocks Further Experimental Analysis Discussion, Reproducibility Statement, Acknowledgements and References A Duality Between Downweighted Residual and Restricting Updates In Linear Layers B Block Layouts C Additional Experiments D Implementation Details 2 RELATED WORK Simplifying deep NNs by removing block components has received a lot of attention, both in transformers and other architectures. In these works, signal propagation theory often acts as inspiration. It has been shown that judicious use of weight initialisations and architectural tools, like skip connections and normalisation layers, can improve signal propagation degeneracies and the trainability of deep NNs. Such considerations have motivated principled modifications with simpler architectures. De & Smith (2020) show that an implicit mechanism of Pre-LN skip connections is to downweight the residual branch relative to the skip branch, leading to better signal propagation. They also show that explicitly downweighting the residual branch allows normalisation layers to be removed without affecting performance. The idea of downweighting residuals for improved signal propagation & trainability has been studied extensively in the literature (Zhang et al., 2018; Hanin & Rolnick, 2018; Tarnowski et al., 2019; Zhang et al., 2019; Arpit et al., 2019; Xu et al., 2020; Bachlechner et al., 2021; Touvron et al., 2021; Hayou et al., 2021; Hayou & Yang, 2023; Martens et al., 2021; Davis et al., 2021; Noci et al., 2022; Wang et al., 2022a; Huang et al., 2020; Wang et al., 2022b). For skip connections (He et al., 2016), it has been shown that transforming non-linear activation functions in MLPs and CNNs to be more linear according to a given deep architecture can enable good signal propagation even without skip connections (Martens et al., 2021; Zhang et al., 2022; Li et al., 2022). He et al. (2023) apply similar considerations to the self-attention mechanism, where the key insight is that attention matrices need to be more identity-like in order to prevent signal degradation in skipless transformers. However, these works find that skipless architectures suffer from significant losses in training speed compared to their residual counterparts, when using standard optimisers like SGD or Adam. Such differences were not observed with stronger optimisers like K-FAC (Martens & Grosse, 2015) on CNNs, and this inability to explain training phenomena highlights a current limitation of signal propagation theory. Ding et al. (2021; 2023) design a CNN, RepVGG, that can be trained like a residual architecture for fast per-update convergence, but reparameterised to be skipless at test time for significantly higher inference throughput. This reparameterisation is related to our considerations of value and projection parameters in Sec. 4. Many works have considered simplifications or improvements specific to the transformer architecture. Most relevant to our work is the parallel block of Wang & Komatsuzaki (2021) (pictured Fig. 1, bottom right), which computes the MLP and attention sub-blocks in parallel for efficiency gains, with minimal performance loss. Trockman & Kolter (2023) observe that the product of value and projection parameters often has a large identity component in trained transformers, and design an initialisation mimicking this to improve performance in standard transformers on small datasets. We find these matrices can be fixed to the identity without loss of performance, which removes them from our simplified architecture. Other works have considered reducing the frequency of MLP sub-blocks (Sridhar et al., 2022; Pires et al., 2023) or efficient replacements to softmax attention (Katharopoulos et al., 2020; Schlag et al., 2021; Choromanski et al., 2021). Sukhbaatar et al. (2019) remove the MLP by integrating it into the attention sub-block, augmented with persistent memory This paper is available on arxiv under CC 4.0 license. Authors: (1) Bobby He, Department of Computer Science, ETH Zurich (Correspondence to: bobby.he@inf.ethz.ch.); (2) Thomas Hofmann, Department of Computer Science, ETH Zurich. Authors: Authors: (1) Bobby He, Department of Computer Science, ETH Zurich (Correspondence to: bobby.he@inf.ethz.ch.); (2) Thomas Hofmann, Department of Computer Science, ETH Zurich. Table of Links Abstract and Introduction Abstract and Introduction Related Work Related Work Preliminaries Preliminaries Simplifying Transformer Blocks Simplifying Transformer Blocks Further Experimental Analysis Further Experimental Analysis Discussion, Reproducibility Statement, Acknowledgements and References Discussion, Reproducibility Statement, Acknowledgements and References A Duality Between Downweighted Residual and Restricting Updates In Linear Layers A Duality Between Downweighted Residual and Restricting Updates In Linear Layers B Block Layouts B Block Layouts C Additional Experiments C Additional Experiments D Implementation Details D Implementation Details 2 RELATED WORK Simplifying deep NNs by removing block components has received a lot of attention, both in transformers and other architectures. In these works, signal propagation theory often acts as inspiration. It has been shown that judicious use of weight initialisations and architectural tools, like skip connections and normalisation layers, can improve signal propagation degeneracies and the trainability of deep NNs. Such considerations have motivated principled modifications with simpler architectures. De & Smith (2020) show that an implicit mechanism of Pre-LN skip connections is to downweight the residual branch relative to the skip branch, leading to better signal propagation. They also show that explicitly downweighting the residual branch allows normalisation layers to be removed without affecting performance. The idea of downweighting residuals for improved signal propagation & trainability has been studied extensively in the literature (Zhang et al., 2018; Hanin & Rolnick, 2018; Tarnowski et al., 2019; Zhang et al., 2019; Arpit et al., 2019; Xu et al., 2020; Bachlechner et al., 2021; Touvron et al., 2021; Hayou et al., 2021; Hayou & Yang, 2023; Martens et al., 2021; Davis et al., 2021; Noci et al., 2022; Wang et al., 2022a; Huang et al., 2020; Wang et al., 2022b). For skip connections (He et al., 2016), it has been shown that transforming non-linear activation functions in MLPs and CNNs to be more linear according to a given deep architecture can enable good signal propagation even without skip connections (Martens et al., 2021; Zhang et al., 2022; Li et al., 2022). He et al. (2023) apply similar considerations to the self-attention mechanism, where the key insight is that attention matrices need to be more identity-like in order to prevent signal degradation in skipless transformers. However, these works find that skipless architectures suffer from significant losses in training speed compared to their residual counterparts, when using standard optimisers like SGD or Adam. Such differences were not observed with stronger optimisers like K-FAC (Martens & Grosse, 2015) on CNNs, and this inability to explain training phenomena highlights a current limitation of signal propagation theory. Ding et al. (2021; 2023) design a CNN, RepVGG, that can be trained like a residual architecture for fast per-update convergence, but reparameterised to be skipless at test time for significantly higher inference throughput. This reparameterisation is related to our considerations of value and projection parameters in Sec. 4. Many works have considered simplifications or improvements specific to the transformer architecture. Most relevant to our work is the parallel block of Wang & Komatsuzaki (2021) (pictured Fig. 1, bottom right), which computes the MLP and attention sub-blocks in parallel for efficiency gains, with minimal performance loss. Trockman & Kolter (2023) observe that the product of value and projection parameters often has a large identity component in trained transformers, and design an initialisation mimicking this to improve performance in standard transformers on small datasets. We find these matrices can be fixed to the identity without loss of performance, which removes them from our simplified architecture. Other works have considered reducing the frequency of MLP sub-blocks (Sridhar et al., 2022; Pires et al., 2023) or efficient replacements to softmax attention (Katharopoulos et al., 2020; Schlag et al., 2021; Choromanski et al., 2021). Sukhbaatar et al. (2019) remove the MLP by integrating it into the attention sub-block, augmented with persistent memory This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Simplifying Transformer Blocks: Related Work

If you’re looking for a beginner-friendly guide on how to launch your first mobile app, you come to the right place.
The HackerNoon Mobile App was also our first jump into the world of app development and we recognize that at first, we were just as lost as you probably are right now. Let us help you get your app off the ground by sharing our own hard-won lessons. Remember to refer to Apple's and Google's official documentation for detailed step-by-step instructions and guidelines to ensure a successful app submission process. Table of Links A Comprehensive Guide From HackerNoon
Google Play Store Step-by-step Guide
Apple App Store Step-by-step Guide
Release Notes Android Version 2.0 - May 11, 2024 Did you know that 74% have absolutely zero faith in the government’s ability to regulate AI? We’ve brought HackerNoon’s original polling data into The App. This release also added hundreds of story templates to make it easier to start writing and made the headline story function more like Google Keep, editable in the story draft itself and within story settings. Last but also first, we improved and sped up authentication to save you a few precious moments upon opening the HackerNoon App. Release notes prior to Version 2.0 are the same for Android and iOS. Apple Version 2.0 - May 11, 2024 Did you know that 74% of people have absolutely zero faith in the government’s ability to regulate AI? Or that 49% of people think 24 is the best age? We’ve brought HackerNoon’s original polling data, and active technology polls, into The HackerNoon App to help you stay on the pulse of the internet’s trends. This release also includes major improvements to the mobile writing experience. We added hundreds of story templates to make it easier to start writing and made the headline story function more like Apple Notes, editable in the story draft itself and within story settings. Breakthrough ideas can come from tiny keyboards! Last but also first, we improved and sped up authentication to save you a few precious moments upon opening the HackerNoon App. Keep the feedback coming! We’ll keep improving! Version 1.9 - Feb 13, 2024 Text editor, now live! You can write stories and submit them directly to human editors via the HackerNoon mobile app. Long live those who write! And don’t forget those who read. We added 12 new languages for HackerNoon stories: Spanish, Hindi, Mandarin, Vietnamese, French, Portuguese, Japanese, Russian, Korean, Turkish, Bengali, and German. We redid how we measure trending tags and publishing activity to measure and display trending technology categories. Also, we added original pixelated icons, dark mode details, a modernized draggable playlist, and redesigned the story cards to better highlight top technology stories. Version 1.8 - Oct 18, 2023 UX Details Galore. From comments display to playlist interactions to story scrolling, this app is smoother and tighter than it was last week. Again, big thanks to the HackerNoon community for providing the detailed feedback. Also the technology categories, like #programming, #ai, and #web3, now have delightful banner images because we are dedicated to making the story more welcoming. And omg, did you see our ranking on the US news chart this week 🤗🥺 Version 1.7 - Sep 25, 2023 A Multilingual One Word App Toggles. TL;DR we added the TL;DR for every story, so if lazy, just read that. More HackerNoon core updates are: Changed name from "HackerNoon App" to just "HackerNoon" under the assumption that you already know this to be a mobile application, which may or may not predict the future.
Published all story translation pages from HackerNoon CMS, so all top and trending stories are now also available in Spanish, Hindi, Mandarin, Vietnamese, French, Portuguese, and Japanese.
Added a settings toggle to hide/show the headphone icon. Read or Listen. Either way you're learning technology distraction-free.
Fixed tags pages to show stories by date with proper story counts. Wonky data be gone.
Fixed typos. Unfortunately "becoming" read as "becoming" in the v1.6. Thank you HackerNoon community for the feedback. Version 1.2 - Sep 11, 2023 Some fixes, push notifications, universal link, improved navigation. If you’re looking for a beginner-friendly guide on how to launch your first mobile app, you come to the right place. The HackerNoon Mobile App was also our first jump into the world of app development and we recognize that at first, we were just as lost as you probably are right now. Let us help you get your app off the ground by sharing our own hard-won lessons. If you’re looking for a beginner-friendly guide on how to launch your first mobile app, you come to the right place. The HackerNoon Mobile App was also our first jump into the world of app development and we recognize that at first, we were just as lost as you probably are right now. Let us help you get your app off the ground by sharing our own hard-won lessons. HackerNoon Mobile App Remember to refer to Apple's and Google's official documentation for detailed step-by-step instructions and guidelines to ensure a successful app submission process. Remember to refer to Apple's and Google's official documentation for detailed step-by-step instructions and guidelines to ensure a successful app submission process. Remember to refer to Apple's and Google's official documentation for detailed step-by-step instructions and guidelines to ensure a successful app submission process. Table of Links A Comprehensive Guide From HackerNoon Google Play Store Step-by-step Guide Apple App Store Step-by-step Guide Release Notes A Comprehensive Guide From HackerNoon A Comprehensive Guide From HackerNoon Google Play Store Step-by-step Guide Google Play Store Step-by-step Guide Apple App Store Step-by-step Guide Apple App Store Step-by-step Guide Release Notes Release Notes Android Android Version 2.0 - May 11, 2024 Version 2.0 - May 11, 2024 Did you know that 74% have absolutely zero faith in the government’s ability to regulate AI? We’ve brought HackerNoon’s original polling data into The App. This release also added hundreds of story templates to make it easier to start writing and made the headline story function more like Google Keep, editable in the story draft itself and within story settings. Last but also first, we improved and sped up authentication to save you a few precious moments upon opening the HackerNoon App. Release notes prior to Version 2.0 are the same for Android and iOS. Release notes prior to Version 2.0 are the same for Android and iOS. Apple Apple Version 2.0 - May 11, 2024 Version 2.0 - May 11, 2024 Did you know that 74% of people have absolutely zero faith in the government’s ability to regulate AI? Or that 49% of people think 24 is the best age? We’ve brought HackerNoon’s original polling data, and active technology polls, into The HackerNoon App to help you stay on the pulse of the internet’s trends. This release also includes major improvements to the mobile writing experience. We added hundreds of story templates to make it easier to start writing and made the headline story function more like Apple Notes, editable in the story draft itself and within story settings. Breakthrough ideas can come from tiny keyboards! Last but also first, we improved and sped up authentication to save you a few precious moments upon opening the HackerNoon App. Keep the feedback coming! We’ll keep improving! Version 1.9 - Feb 13, 2024 Version 1.9 - Feb 13, 2024 Text editor, now live! You can write stories and submit them directly to human editors via the HackerNoon mobile app. Long live those who write! And don’t forget those who read. We added 12 new languages for HackerNoon stories: Spanish, Hindi, Mandarin, Vietnamese, French, Portuguese, Japanese, Russian, Korean, Turkish, Bengali, and German. We redid how we measure trending tags and publishing activity to measure and display trending technology categories. Also, we added original pixelated icons, dark mode details, a modernized draggable playlist, and redesigned the story cards to better highlight top technology stories. Version 1.8 - Oct 18, 2023 Version 1.8 - Oct 18, 2023 UX Details Galore. From comments display to playlist interactions to story scrolling, this app is smoother and tighter than it was last week. Again, big thanks to the HackerNoon community for providing the detailed feedback. Also the technology categories, like #programming, #ai, and #web3, now have delightful banner images because we are dedicated to making the story more welcoming. And omg, did you see our ranking on the US news chart this week 🤗🥺 Version 1.7 - Sep 25, 2023 Version 1.7 - Sep 25, 2023 A Multilingual One Word App Toggles. TL;DR we added the TL;DR for every story, so if lazy, just read that. More HackerNoon core updates are: Changed name from "HackerNoon App" to just "HackerNoon" under the assumption that you already know this to be a mobile application, which may or may not predict the future. Published all story translation pages from HackerNoon CMS, so all top and trending stories are now also available in Spanish, Hindi, Mandarin, Vietnamese, French, Portuguese, and Japanese. Added a settings toggle to hide/show the headphone icon. Read or Listen. Either way you're learning technology distraction-free. Fixed tags pages to show stories by date with proper story counts. Wonky data be gone. Fixed typos. Unfortunately "becoming" read as "becoming" in the v1.6. Thank you HackerNoon community for the feedback. Changed name from "HackerNoon App" to just "HackerNoon" under the assumption that you already know this to be a mobile application, which may or may not predict the future. Published all story translation pages from HackerNoon CMS, so all top and trending stories are now also available in Spanish, Hindi, Mandarin, Vietnamese, French, Portuguese, and Japanese. Added a settings toggle to hide/show the headphone icon. Read or Listen. Either way you're learning technology distraction-free. Fixed tags pages to show stories by date with proper story counts. Wonky data be gone. Fixed typos. Unfortunately "becoming" read as "becoming" in the v1.6. Thank you HackerNoon community for the feedback. Version 1.2 - Sep 11, 2023 Version 1.2 - Sep 11, 2023 Some fixes, push notifications, universal link, improved navigation.

How to Launch an App: HackerNoon Release Notes

কীভাবে একটি অ্যাপ চালু করবেন: হ্যাকারনুন রিলিজ নোট

So starten Sie eine App: Versionshinweise zu HackerNoon

Cómo iniciar una aplicación: Notas de la versión de HackerNoon

Comment lancer une application : notes de version de HackerNoon

ऐप कैसे लॉन्च करें: हैकरनून रिलीज़ नोट्स

アプリの起動方法: HackerNoon リリースノート

앱 실행 방법: HackerNoon 릴리스 노트

Como iniciar um aplicativo: Notas de versão do HackerNoon

Как запустить приложение: примечания к выпуску HackerNoon

Bir Uygulama Nasıl Başlatılır: HackerNoon Sürüm Notları

Cách khởi chạy ứng dụng: Ghi chú phát hành HackerNoon

如何启动应用程序：HackerNoon 发行说明

Authors:
(1) Mohammadreza Hazhirpasand, University of Bern, Bern, Switzerland;
(2) Oscar Nierstrasz, University of Bern, Bern, Switzerland;
(3) Mohammad Ghafari, University of Auckland, Auckland, New Zealand. Table of Links Abstract and I. Introduction
II. Methodology
III. Results and Discussion
IV. Threats to Validity
V. Related Work
VI. Conclusions
VII. Acknowledgments and References V. RELATED WORK Kafader and Ghafari developed FluentCrypto with the goal of creating usable and secure crypto APIs for developers [13]. FluentCrypto hides the low-level complexities that involve using a native API and provides a task-based solution that novices can use without crypto knowledge. It also allows crypto experts to configure the API as they find fit and uses a set of pre-defined rules to determine the configuration is secure. Green et al. proposed ten principles to aid library developers in reducing the possibility of API misuses [3]. For instance, one of the principles is to make defaults safe and unambiguous in APIs. This principle can significantly lessen the hardship of developers as we have witnessed their confusion about default values in library interoperability and encryption/decryption themes. Reviewing 2491 Stack Overflow questions in relation to seven crypto libraries, Patnaik et al. identified 16 underlying usability issues in crypto libraries [6]. There is a common theme between their work and this study, which is the demand for example code snippets. However, their objective was to investigate the usability of crypto APIs, whereas we grouped the problems into themes based on the technical perspective. Developers refer to Stack Overflow as popular documentation. Parnin et al. studied three popular non-crypto APIs (Android, GWT, Java) to observe the quality and dynamics of the Stack Overflow documentation for these APIs [14]. They found that the crowd generated a rich source of content containing code examples, which are viewed by a majority of developers. For instance, more than 35 000 developers contributed to Android API discussions, which covers 87% of the classes and has been viewed over 80 million times. However, with the massive number of discussions, there is a small pool of experts available to answer the questions. Hou et al. conducted a manual analysis on a set of newsgroup discussions to understand developer problems in using APIs [15]. They described 15 obstacles, e.g., unclear API semantics by design or wrong parameter values, which hinder developers, and alleviating such obstacles increases the accessibility of APIs. Hazhirpasand et al. conducted a large-scale study on crypto-related posts on Stack Overflow by using Latent Dirichlet Allocation (LDA), which is a generative statistical model, to cluster 91 954 questions [4]. They found three high-level themes in developers’ questions, namely digital certificate, programming issues, and password/hashing. In contrast, we reported more elaborated themes of issues specifically related to crypto libraries and excluded configurational problems and general-purpose crypto questions. A recent study analyzed 489 Java projects in which the majority of APIs (i.e., 13 of 15) were misused at least once [16]. In addition, contacting the developers showed that security warnings in the documentation of crypto APIs are rare. Consequently, based on a recent survey, developers incline to resolve their crypto issues on Stack Overflow, where the authenticity of the provided answers is debatable [17]. This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Mohammadreza Hazhirpasand, University of Bern, Bern, Switzerland; (2) Oscar Nierstrasz, University of Bern, Bern, Switzerland; (3) Mohammad Ghafari, University of Auckland, Auckland, New Zealand. Authors: Authors: (1) Mohammadreza Hazhirpasand, University of Bern, Bern, Switzerland; (2) Oscar Nierstrasz, University of Bern, Bern, Switzerland; (3) Mohammad Ghafari, University of Auckland, Auckland, New Zealand. Table of Links Abstract and I. Introduction II. Methodology III. Results and Discussion IV. Threats to Validity V. Related Work VI. Conclusions VII. Acknowledgments and References Abstract and I. Introduction Abstract and I. Introduction II. Methodology II. Methodology III. Results and Discussion III. Results and Discussion IV. Threats to Validity IV. Threats to Validity V. Related Work V. Related Work VI. Conclusions VI. Conclusions VII. Acknowledgments and References VII. Acknowledgments and References V. RELATED WORK Kafader and Ghafari developed FluentCrypto with the goal of creating usable and secure crypto APIs for developers [13]. FluentCrypto hides the low-level complexities that involve using a native API and provides a task-based solution that novices can use without crypto knowledge. It also allows crypto experts to configure the API as they find fit and uses a set of pre-defined rules to determine the configuration is secure. Green et al. proposed ten principles to aid library developers in reducing the possibility of API misuses [3]. For instance, one of the principles is to make defaults safe and unambiguous in APIs. This principle can significantly lessen the hardship of developers as we have witnessed their confusion about default values in library interoperability and encryption/decryption themes. Reviewing 2491 Stack Overflow questions in relation to seven crypto libraries, Patnaik et al. identified 16 underlying usability issues in crypto libraries [6]. There is a common theme between their work and this study, which is the demand for example code snippets. However, their objective was to investigate the usability of crypto APIs, whereas we grouped the problems into themes based on the technical perspective. Developers refer to Stack Overflow as popular documentation. Parnin et al. studied three popular non-crypto APIs (Android, GWT, Java) to observe the quality and dynamics of the Stack Overflow documentation for these APIs [14]. They found that the crowd generated a rich source of content containing code examples, which are viewed by a majority of developers. For instance, more than 35 000 developers contributed to Android API discussions, which covers 87% of the classes and has been viewed over 80 million times. However, with the massive number of discussions, there is a small pool of experts available to answer the questions. Hou et al. conducted a manual analysis on a set of newsgroup discussions to understand developer problems in using APIs [15]. They described 15 obstacles, e.g., unclear API semantics by design or wrong parameter values, which hinder developers, and alleviating such obstacles increases the accessibility of APIs. Hazhirpasand et al. conducted a large-scale study on crypto-related posts on Stack Overflow by using Latent Dirichlet Allocation (LDA), which is a generative statistical model, to cluster 91 954 questions [4]. They found three high-level themes in developers’ questions, namely digital certificate, programming issues, and password/hashing. In contrast, we reported more elaborated themes of issues specifically related to crypto libraries and excluded configurational problems and general-purpose crypto questions. A recent study analyzed 489 Java projects in which the majority of APIs (i.e., 13 of 15) were misused at least once [16]. In addition, contacting the developers showed that security warnings in the documentation of crypto APIs are rare. Consequently, based on a recent survey, developers incline to resolve their crypto issues on Stack Overflow, where the authenticity of the provided answers is debatable [17]. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

Dazed and Confused: What’s Wrong with Crypto
Libraries? — Related Work

Authors:
(1) Aarav Patel, Amity Regional High School – email: aarav.dhp@gmail.com;
(2) Peter Gloor, Center for Collective Intelligence, Massachusetts Institute of Technology and Corresponding author – email: pgloor@mit.edu. Table of Links Abstract and Introduction
Related Works
Purpose
Methods
Results
Discussion
Conclusion and Bibliography 2. Related Works Existing ESG-related research falls under two main categories. Some papers aim to correlate ESG performance with financial performance and see if a company’s Corporate Social Responsibility (CSR) can be used to predict future stock performance (Jain et al., 2019). Other papers propose new data-driven methods for enhancing and automating ESG rating measurement to avoid existing fallacies/inefficiencies (Hisano et al., 2020; Krappel et al., 2021; Liao et al., 2017; Lin et al., 2018; Shahi et al., 2011; Sokolov et al., 2021; Venturelli et al., 2017; Wicher et al., 2019). This paper will fall into the latter category. Since many firms publish sustainability reports on an annual basis, many researchers use this content for analysis. This is typically done using text mining to identify ESG topics and trends. In order to parse out and leverage this data, researchers have created classification models that can classify sentences/paragraphs into various ESG subdimensions (Liao et al., 2017; Lin et al., 2018). Additionally, some researchers have used these text classification algorithms to analyze the completeness of sustainability reports (Shahi et al., 2011). This is because companies sometimes limit disclosure regarding negative ESG aspects within their filings. Both tools can assist in automatic ESG scoring using company filings, which increases access for companies without ESG coverage. However, there are deficiencies in solely relying on self-reported filings for analysis since it fails to consider omitted data or newer developments. As a result, researchers have been testing out alternative methods to solve this. For instance, some researchers utilize Fuzzy Expert System (FES) or a Fuzzy Analytic Network Process (FANP), pulling data from quantitative indicators (i.e., metrics provided by the Global Reporting Initiative) and qualitative features from surveys/interviews (Venturelli et al., 2017; Wicher et al., 2019). Others collected data from online social networks like Twitter to analyze a company’s sustainability profile. For example, some used Natural Language Processing (NLP) frameworks to classify Tweets into various ESG topics and determine whether they are positive or negative (Sokolov et al., 2021). Furthermore, some used heterogeneous information networks that combined data from various negative news datasets and used machine learning to predict ESG (Hisano et al., 2020). Finally, others explored the viability of using fundamental data such as a company’s profile and financials to predict ESG (Krappel et al., 2021). Overall, all these methods aimed to improve self-reported filings by using more balanced, unbiased, and real-time data. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: (1) Aarav Patel, Amity Regional High School – email: aarav.dhp@gmail.com; (2) Peter Gloor, Center for Collective Intelligence, Massachusetts Institute of Technology and Corresponding author – email: pgloor@mit.edu. Authors: Authors: (1) Aarav Patel, Amity Regional High School – email: aarav.dhp@gmail.com; (2) Peter Gloor, Center for Collective Intelligence, Massachusetts Institute of Technology and Corresponding author – email: pgloor@mit.edu. Table of Links Abstract and Introduction Related Works Purpose Methods Results Discussion Conclusion and Bibliography Abstract and Introduction Abstract and Introduction Related Works Related Works Purpose Purpose Methods Methods Results Results Discussion Discussion Conclusion and Bibliography Conclusion and Bibliography 2. Related Works Existing ESG-related research falls under two main categories. Some papers aim to correlate ESG performance with financial performance and see if a company’s Corporate Social Responsibility (CSR) can be used to predict future stock performance (Jain et al., 2019). Other papers propose new data-driven methods for enhancing and automating ESG rating measurement to avoid existing fallacies/inefficiencies (Hisano et al., 2020; Krappel et al., 2021; Liao et al., 2017; Lin et al., 2018; Shahi et al., 2011; Sokolov et al., 2021; Venturelli et al., 2017; Wicher et al., 2019). This paper will fall into the latter category. Since many firms publish sustainability reports on an annual basis, many researchers use this content for analysis. This is typically done using text mining to identify ESG topics and trends. In order to parse out and leverage this data, researchers have created classification models that can classify sentences/paragraphs into various ESG subdimensions (Liao et al., 2017; Lin et al., 2018). Additionally, some researchers have used these text classification algorithms to analyze the completeness of sustainability reports (Shahi et al., 2011). This is because companies sometimes limit disclosure regarding negative ESG aspects within their filings. Both tools can assist in automatic ESG scoring using company filings, which increases access for companies without ESG coverage. However, there are deficiencies in solely relying on self-reported filings for analysis since it fails to consider omitted data or newer developments. As a result, researchers have been testing out alternative methods to solve this. For instance, some researchers utilize Fuzzy Expert System (FES) or a Fuzzy Analytic Network Process (FANP), pulling data from quantitative indicators (i.e., metrics provided by the Global Reporting Initiative) and qualitative features from surveys/interviews (Venturelli et al., 2017; Wicher et al., 2019). Others collected data from online social networks like Twitter to analyze a company’s sustainability profile. For example, some used Natural Language Processing (NLP) frameworks to classify Tweets into various ESG topics and determine whether they are positive or negative (Sokolov et al., 2021). Furthermore, some used heterogeneous information networks that combined data from various negative news datasets and used machine learning to predict ESG (Hisano et al., 2020). Finally, others explored the viability of using fundamental data such as a company’s profile and financials to predict ESG (Krappel et al., 2021). Overall, all these methods aimed to improve self-reported filings by using more balanced, unbiased, and real-time data. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. available on arxiv

Creating a Systematic ESG Scoring System: Related Works

একটি পদ্ধতিগত ESG স্কোরিং সিস্টেম তৈরি করা: সম্পর্কিত কাজ

Erstellen eines systematischen ESG-Bewertungssystems: Verwandte Arbeiten

Creación de un sistema sistemático de puntuación ESG: trabajos relacionados

Créer un système de notation ESG systématique : travaux connexes

एक व्यवस्थित ईएसजी स्कोरिंग प्रणाली बनाना: संबंधित कार्य

体系的なESGスコアリングシステムの構築：関連研究

체계적인 ESG 채점 시스템 구축: 관련 연구

Criando um Sistema Sistemático de Pontuação ESG: Trabalhos Relacionados

Создание систематической системы оценки ESG: сопутствующие работы

Sistematik ÇSY Puanlama Sistemi Oluşturma: İlgili Çalışmalar

Tạo hệ thống tính điểm ESG có hệ thống: Công việc liên quan

创建系统的 ESG 评分体系：相关工作

Authors:
(1) P Aditya Sreekar, Amazon and these authors contributed equally to this work {sreekarp@amazon.com};
(2) Sahil Verm, Amazon and these authors contributed equally to this work {vrsahil@amazon.com;}
(3) Varun Madhavan, Indian Institute of Technology, Kharagpur. Work done during internship at Amazon {varunmadhavan@iitkgp.ac.in};
(4) Abhishek Persad, Amazon {persadap@amazon.com}. Table of Links Abstract and Introduction
Related Works
Methodology
Experiment
Conclusion and Future Work
References 2. Related Works Tree-based algorithms are widely used in machine learning for tabular data. Decision trees recursively split the data into multiple parts based on axis-aligned hyper-planes (Hastie et al., 2009). Random Forests (RF) (Breiman, 2001) and Gradient Boosted Decision Trees (GBDT) (Friedman, 2001) are the most commonly used tree based ensembles. RF fits multiple decision trees on random subsets of the data and averages/polls the predictions to alleviate the overfitting characteristic of decision trees. GBDT, XGBoost (Chen and Guestrin, 2016), and CatBoost (Prokhorenkova et al., 2018) are boosted ensemble models that sequentially build decision trees to correct errors made by previous trees, leading to improved performance on complex datasets with non-linear relations. Recently, there has been a lot of interest in deep learning models for tabular data. Some methods introduce differentiable approximations of decision functions used in decision trees to make them differentiable (Hazimeh et al., 2020; Popov et al., 2019). These methods outperform pure tree based problem for some problem statements, however, they are not consistently better (Gorishniy et al., 2021). Other methods have used attention mechanisms to adapt DL methods to tabular data (Arik et al., 2019; Huang et al., 2020; Gorishniy et al., 2021; Somepalli et al., 2021; Chen et al., 2022). TabNet (Arik et al., 2019) proposes a sparse attention mechanism that is stacked in multiple layers to mimic the recursive splitting of decision trees. Inspired from the success of self-attention transformers (Vaswani et al., 2017) in many domains (Devlin et al., 2019; Dosovitskiy et al., 2021; Gong et al., 2021) methods like TabTransformer (Huang et al., 2020), FT-Transformer (Gorishniy et al., 2021) and SAINT (Somepalli et al., 2021) were proposed. TabTransformer embeds all categorical variables into a unified embedding space, and a sentence of categorical embeddings is passed through self-attention transformer layers. FT-Transformer further extends this by attending to numerical features as well, by using continuous embedding. SAINT builds on FT-Transformer by proposing a new kind of attention which captures interactions between samples of a batch. However, SAINT does not provide any advantage over FT-Transformer for our problem statement, because intersample attention is only effective when the number of dimensions is higher in comparision to the number of samples, thus we do not compare RCT against SAINT (Somepalli et al., 2021). This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: (1) P Aditya Sreekar, Amazon and these authors contributed equally to this work {sreekarp@amazon.com}; (2) Sahil Verm, Amazon and these authors contributed equally to this work {vrsahil@amazon.com;} (3) Varun Madhavan, Indian Institute of Technology, Kharagpur. Work done during internship at Amazon {varunmadhavan@iitkgp.ac.in}; (4) Abhishek Persad, Amazon {persadap@amazon.com}. Authors: Authors: (1) P Aditya Sreekar, Amazon and these authors contributed equally to this work {sreekarp@amazon.com}; (2) Sahil Verm, Amazon and these authors contributed equally to this work {vrsahil@amazon.com;} (3) Varun Madhavan, Indian Institute of Technology, Kharagpur. Work done during internship at Amazon {varunmadhavan@iitkgp.ac.in}; (4) Abhishek Persad, Amazon {persadap@amazon.com}. Table of Links Abstract and Introduction Related Works Methodology Experiment Conclusion and Future Work References Abstract and Introduction Abstract and Introduction Related Works Related Works Methodology Methodology Experiment Experiment Conclusion and Future Work Conclusion and Future Work References References 2. Related Works Tree-based algorithms are widely used in machine learning for tabular data. Decision trees recursively split the data into multiple parts based on axis-aligned hyper-planes (Hastie et al., 2009). Random Forests (RF) (Breiman, 2001) and Gradient Boosted Decision Trees (GBDT) (Friedman, 2001) are the most commonly used tree based ensembles. RF fits multiple decision trees on random subsets of the data and averages/polls the predictions to alleviate the overfitting characteristic of decision trees. GBDT, XGBoost (Chen and Guestrin, 2016), and CatBoost (Prokhorenkova et al., 2018) are boosted ensemble models that sequentially build decision trees to correct errors made by previous trees, leading to improved performance on complex datasets with non-linear relations. Recently, there has been a lot of interest in deep learning models for tabular data. Some methods introduce differentiable approximations of decision functions used in decision trees to make them differentiable (Hazimeh et al., 2020; Popov et al., 2019). These methods outperform pure tree based problem for some problem statements, however, they are not consistently better (Gorishniy et al., 2021). Other methods have used attention mechanisms to adapt DL methods to tabular data (Arik et al., 2019; Huang et al., 2020; Gorishniy et al., 2021; Somepalli et al., 2021; Chen et al., 2022). TabNet (Arik et al., 2019) proposes a sparse attention mechanism that is stacked in multiple layers to mimic the recursive splitting of decision trees. Inspired from the success of self-attention transformers (Vaswani et al., 2017) in many domains (Devlin et al., 2019; Dosovitskiy et al., 2021; Gong et al., 2021) methods like TabTransformer (Huang et al., 2020), FT-Transformer (Gorishniy et al., 2021) and SAINT (Somepalli et al., 2021) were proposed. TabTransformer embeds all categorical variables into a unified embedding space, and a sentence of categorical embeddings is passed through self-attention transformer layers. FT-Transformer further extends this by attending to numerical features as well, by using continuous embedding. SAINT builds on FT-Transformer by proposing a new kind of attention which captures interactions between samples of a batch. However, SAINT does not provide any advantage over FT-Transformer for our problem statement, because intersample attention is only effective when the number of dimensions is higher in comparision to the number of samples, thus we do not compare RCT against SAINT (Somepalli et al., 2021). This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. available on arxiv

Unveiling the Power of Self-Attention for Shipping Cost Prediction: Related Works

শিপিং খরচ পূর্বাভাসের জন্য স্ব-মনোযোগের শক্তি উন্মোচন: সম্পর্কিত কাজ

Die Macht der Selbstaufmerksamkeit zur Vorhersage von Versandkosten enthüllen: Verwandte Arbeiten

Revelando el poder de la autoatención para la predicción de costos de envío: trabajos relacionados

Dévoiler le pouvoir de l’auto-attention pour la prévision des coûts d’expédition : travaux connexes

शिपिंग लागत की भविष्यवाणी के लिए स्व-ध्यान की शक्ति का अनावरण: संबंधित कार्य

配送コスト予測における自己注意の力を明らかにする: 関連研究

배송비 예측을 위한 자기주의의 힘 공개: 관련 연구

Revelando o poder da autoatenção para previsão de custos de envio: trabalhos relacionados

Раскрытие силы самовнимания для прогнозирования стоимости доставки: сопутствующие работы

Nakliye Maliyeti Tahmininde Kişisel Dikkatin Gücünü Ortaya Çıkarma: İlgili Çalışmalar

Khám phá sức mạnh của sự chú ý trong việc dự đoán chi phí vận chuyển: Các tác phẩm liên quan

揭示自注意力机制在运输成本预测中的作用：相关研究

Authors:
(1) Simon R. Davies, School of Computing, Edinburgh Napier University, Edinburgh, UK (s.davies@napier.ac.uk);
(2) Richard Macfarlane, School of Computing, Edinburgh Napier University, Edinburgh, UK;
(3) William J. Buchanan, School of Computing, Edinburgh Napier University, Edinburgh, UK. Table of Links Abstract and 1 Introduction
2. Related Work
3. Methodology and 3.1. File Content Analysis
3.2. File Name Analysis
3.3. Executable Analysis
3.4. Behaviour Analysis
4. Evaluation and Discussion
4.1. Majority Voting
5. Conclusion
5.1. Limitations
5.2. Future Work
References and Appendix 2. Related Work Over the last 20 years, a significant number of ransomware detection systems have been proposed in the research literature. The approaches used by these detection systems can be loosely divided into two categories. In one approach, a single method or test is developed which is then used to determine if the system is being attacked by ransomware. The alternative approach is to use machine learning to perform the identification. With the machine learning approach, the system designers identify key features from the running process and system under investigation. The machine learning model then attempts to determine patterns within these features on which to base its judgement. A decision, or classification, is then made, based on the measured values of these features, as to whether the system is under attack or not. Examples of single-method approaches are [2, 40, 23, 54, 63, 70]. In these cases, the entire effectiveness of the detection technique relies solely on the ability of this single criterion to distinguish between benign and malicious programs [20]. For example, one particular technique used in the identification of ransomware execution is to use the calculated entropy value of the files created by a process. Encrypted files tend to have a high entropy value whereas the entropy value of plain text files is much lower. Encrypted output files generated during the execution of a ransomware program would tend to have higher entropy values, possibly allowing them to be identified as a product of a ransomware infection. Unfortunately, this technique struggles to correctly distinguish between encrypted files and benign files that also have high entropy such as compressed files. The use of entropy as a detection metric has also been called into question [56, 44] as there exist techniques that could be used by ransomware to avoid detection via encoding or, in some other way, manipulating the encrypted output file. Examples of ransomware detection techniques that have leveraged machine learning are [77, 2, 3, 40, 65, 67] or similarly neural networks [5, 31, 50, 55]. These systems are trained using extracted features from typical ransomware processes or systems that are being attacked by ransomware. Examples of features that are used in these systems are: write entropy, file overwrite behaviour, directory traversal, directory listing, cross-file type access, read/write/ create/close operations, temporary files, file type coverage, file similarity, file type change and access frequency [20]. In most cases, with systems that rely on machine learning to determine if a system is being attacked, the significance of the individual extracted features and their subsequent impact on the final classification is represented internally by the detection system’s model and is not immediately obvious to an observer. Inadequacies with this approach have been investigated in the literature [20] which discusses classifier evasion techniques, known as adversarial machine learning that can be leveraged by ransomware developers to avoid classification and subsequent detection. However, in a few proposed ransomware detection systems, the designers do try to provide insight into the machine learning techniques used and how the tested features affect the overall decision-making process. The developers of the detection system UNVEIL [36] and its successor Redemption [37], introduce the concept of a malice score which is a combined weighted score derived from the outcome of individual feature tests. The system detects suspicious activity using dynamic analysis and generates a malice score using a heuristic function. Inputs to this function are various behavioural features such as file entropy changes, writes that cover extended portions of a file, file deletion, processes writing to a large number of user files, processes writing to files of different types and back-to-back writes. CryptoLock [68] propose a similar approach summing the results of various tests into a cumulative scoring they refer to as a Reputation Score. This score is derived from measurements of file type changes, the similarity between original and written content and output file entropy values. Another detection system, RWGuard [57], does mention the specific features that are inspected and include file IO, decoy files, file change monitoring and crypto API monitoring. However, very little detail on how the specific calculations are performed is provided. DNA-Droid [26], was the only detection system found that, leveraged a combination of static and dynamic analysis as the inputs to their neural network model. In all cases if this cumulative score is above a certain threshold, then the process is deemed to be malicious, otherwise, the process is considered benign. However, in all these cases, the individual test results and thresholds are still determined heuristically via the machinelearning model. The model itself decides the significance and weighting given to each extracted feature and the influence that each feature has on the final classification. Reducing the entire decision-making process to effectively a black box function. A consequence of this is that it is difficult for the designers to directly affect the final decision, thus preventing them from being easily able to tune and influence the decision-making process and final classification produced by the model. The resulting quality and accuracy of the decisions made by these systems are essentially reliant on the quality of the training data used to develop the models in the first place. No ransomware detection systems have been identified in the literature that uses a malice scoring type approach where the constituent scores contributing to the final malice score are determined using analytical or algorithmic calculation methods as opposed to the heuristics used in machine learning approaches. This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Simon R. Davies, School of Computing, Edinburgh Napier University, Edinburgh, UK (s.davies@napier.ac.uk); (2) Richard Macfarlane, School of Computing, Edinburgh Napier University, Edinburgh, UK; (3) William J. Buchanan, School of Computing, Edinburgh Napier University, Edinburgh, UK. Authors: Authors: (1) Simon R. Davies, School of Computing, Edinburgh Napier University, Edinburgh, UK (s.davies@napier.ac.uk); (2) Richard Macfarlane, School of Computing, Edinburgh Napier University, Edinburgh, UK; (3) William J. Buchanan, School of Computing, Edinburgh Napier University, Edinburgh, UK. Table of Links Abstract and 1 Introduction 2. Related Work 3. Methodology and 3.1. File Content Analysis 3.2. File Name Analysis 3.3. Executable Analysis 3.4. Behaviour Analysis 4. Evaluation and Discussion 4.1. Majority Voting 5. Conclusion 5.1. Limitations 5.2. Future Work References and Appendix Abstract and 1 Introduction Abstract and 1 Introduction 2. Related Work 2. Related Work 3. Methodology and 3.1. File Content Analysis 3. Methodology and 3.1. File Content Analysis 3.2. File Name Analysis 3.2. File Name Analysis 3.3. Executable Analysis 3.3. Executable Analysis 3.4. Behaviour Analysis 3.4. Behaviour Analysis 4. Evaluation and Discussion 4. Evaluation and Discussion 4.1. Majority Voting 4.1. Majority Voting 5. Conclusion 5. Conclusion 5.1. Limitations 5.1. Limitations 5.2. Future Work 5.2. Future Work References and Appendix References and Appendix 2. Related Work Over the last 20 years, a significant number of ransomware detection systems have been proposed in the research literature. The approaches used by these detection systems can be loosely divided into two categories. In one approach, a single method or test is developed which is then used to determine if the system is being attacked by ransomware. The alternative approach is to use machine learning to perform the identification. With the machine learning approach, the system designers identify key features from the running process and system under investigation. The machine learning model then attempts to determine patterns within these features on which to base its judgement. A decision, or classification, is then made, based on the measured values of these features, as to whether the system is under attack or not. Examples of single-method approaches are [2, 40, 23, 54, 63, 70]. In these cases, the entire effectiveness of the detection technique relies solely on the ability of this single criterion to distinguish between benign and malicious programs [20]. For example, one particular technique used in the identification of ransomware execution is to use the calculated entropy value of the files created by a process. Encrypted files tend to have a high entropy value whereas the entropy value of plain text files is much lower. Encrypted output files generated during the execution of a ransomware program would tend to have higher entropy values, possibly allowing them to be identified as a product of a ransomware infection. Unfortunately, this technique struggles to correctly distinguish between encrypted files and benign files that also have high entropy such as compressed files. The use of entropy as a detection metric has also been called into question [56, 44] as there exist techniques that could be used by ransomware to avoid detection via encoding or, in some other way, manipulating the encrypted output file. Examples of ransomware detection techniques that have leveraged machine learning are [77, 2, 3, 40, 65, 67] or similarly neural networks [5, 31, 50, 55]. These systems are trained using extracted features from typical ransomware processes or systems that are being attacked by ransomware. Examples of features that are used in these systems are: write entropy, file overwrite behaviour, directory traversal, directory listing, cross-file type access, read/write/ create/close operations, temporary files, file type coverage, file similarity, file type change and access frequency [20]. In most cases, with systems that rely on machine learning to determine if a system is being attacked, the significance of the individual extracted features and their subsequent impact on the final classification is represented internally by the detection system’s model and is not immediately obvious to an observer. Inadequacies with this approach have been investigated in the literature [20] which discusses classifier evasion techniques, known as adversarial machine learning that can be leveraged by ransomware developers to avoid classification and subsequent detection. However, in a few proposed ransomware detection systems, the designers do try to provide insight into the machine learning techniques used and how the tested features affect the overall decision-making process. The developers of the detection system UNVEIL [36] and its successor Redemption [37], introduce the concept of a malice score which is a combined weighted score derived from the outcome of individual feature tests. The system detects suspicious activity using dynamic analysis and generates a malice score using a heuristic function. Inputs to this function are various behavioural features such as file entropy changes, writes that cover extended portions of a file, file deletion, processes writing to a large number of user files, processes writing to files of different types and back-to-back writes. CryptoLock [68] propose a similar approach summing the results of various tests into a cumulative scoring they refer to as a Reputation Score. This score is derived from measurements of file type changes, the similarity between original and written content and output file entropy values. Another detection system, RWGuard [57], does mention the specific features that are inspected and include file IO, decoy files, file change monitoring and crypto API monitoring. However, very little detail on how the specific calculations are performed is provided. DNA-Droid [26], was the only detection system found that, leveraged a combination of static and dynamic analysis as the inputs to their neural network model. In all cases if this cumulative score is above a certain threshold, then the process is deemed to be malicious, otherwise, the process is considered benign. However, in all these cases, the individual test results and thresholds are still determined heuristically via the machinelearning model. The model itself decides the significance and weighting given to each extracted feature and the influence that each feature has on the final classification. Reducing the entire decision-making process to effectively a black box function. A consequence of this is that it is difficult for the designers to directly affect the final decision, thus preventing them from being easily able to tune and influence the decision-making process and final classification produced by the model. The resulting quality and accuracy of the decisions made by these systems are essentially reliant on the quality of the training data used to develop the models in the first place. No ransomware detection systems have been identified in the literature that uses a malice scoring type approach where the constituent scores contributing to the final malice score are determined using analytical or algorithmic calculation methods as opposed to the heuristics used in machine learning approaches. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

Majority Voting Approach to Ransomware Detection: Related Work

**TORONTO, Canada, April 22nd, 2024/GamingWire/--**Parallel Studio, the visionary gaming studio that blends art and technology to create cutting-edge gaming experiences, is excited to launch its award winning trading card game on the Epic Games Store. With a team hailing from the likes of Blizzard, Epic Games, Ubisoft, Riot Games, and Xbox Game Studios, Parallel is on course to make waves in the card gaming world. Parallel is a free-to-play digital trading card game set in a captivating sci-fi universe, where five parallel evolutions of humanity battle for control of Earth 10,000 years in the future. Players collect cards, build 40-card decks combining cards from one of the five factions and the Universal pool, and engage in strategic battles against their peers. There are several points that separate Parallel from other TCGs in the space (ex. Magic the Gathering, Hearthstone, etc). Free to Play: players can play and earn cards with no purchase necessary. On top of that, Parallel compensates players that play for free by automatically entering them into weekly and monthly giveaways every season. This is on top of rewards players can get through in-game tournaments, battle pass, referring friends and winning ranked matches.
Bank Cards for Energy: players strategically choose which cards to bank each turn, gaining energy that allows them to play units, relics, effects and upgrades. Players will find that choosing what card to bank and what card to play is critical, as the right combination can create a powerful winning strategy.
Paragons & Factions: Parallel has 5 unique factions with 15 game changing leaders (Paragons) each with their own active/passive abilities helping players craft a unique play style.
Play At Your Pace: Parallel has different game modes for all types of players. Beginners and casual players can enjoy modes like unranked, AI and rookie. Competitive players can participate in the Ranked Ladder and Tournament Modes to earn exclusive rewards and prizes. Parallel is truly built for TCG veterans and new players to the genre. Founded in 2021, Parallel has already seen quite a bit of success. Despite its TCG being in open beta, the game has already won several game of the year awards. The studio has also announced a prize pool of $500,000 for its 2024 competitive scene. Along with that, several professional players and prominent creators that played Hearthstone have joined Parallel’s ambassador program, priming the game to be the next big TCG. Finally, the studio recently released an expansion set called "Planetfall" that is currently playable. The expansion set added over 100 new cards to the game and the studio plans to launch more expansion sets in the future. Quotes From Parallel "Our journey has been nothing short of extraordinary, and being on Epic marks a significant milestone in our global expansion strategy," says Sascha M, CEO & Co-Founder of Parallel. “Parallel going live on Epic opens us up to millions more players worldwide. This expansion into Epic comes alongside the kickoff of our global competitive series, the Parallel League, which has over $500,000 in prizes in its first year. But we don't plan to stop there. As we roll out more ways to access and play Parallel we also plan to scale our competitive prize pool into millions of dollars next year.” - Franklin Fitch, Head of Partnerships, Parallel Studios About Parallel Studios Parallel Studios is a revolutionary game development company creating the next AAA sci-fi franchise. Current projects include the Parallel TCG, an exciting science fiction card game in which five factions battle for supremacy over a future Earth, and Parallel Colony, an AI-powered idle game in which you play through conversation with your Avatar. Parallel aims to continue breaking new ground and bringing amazing experiences to this new IP. Contact Head of Growth Swish Goswami Parallel Studios swish@parallel.game 3472579158 This story was distributed as a release by Gamingwire under HackerNoon’s Business Blogging Program. Learn more about the program here. **TORONTO, Canada, April 22nd, 2024/GamingWire/--**Parallel Studio, the visionary gaming studio that blends art and technology to create cutting-edge gaming experiences, is excited to launch its award winning trading card game on the Epic Games Store. With a team hailing from the likes of Blizzard, Epic Games, Ubisoft, Riot Games, and Xbox Game Studios, Parallel is on course to make waves in the card gaming world. Parallel is a free-to-play digital trading card game set in a captivating sci-fi universe, where five parallel evolutions of humanity battle for control of Earth 10,000 years in the future. Players collect cards, build 40-card decks combining cards from one of the five factions and the Universal pool, and engage in strategic battles against their peers. There are several points that separate Parallel from other TCGs in the space (ex. Magic the Gathering, Hearthstone, etc). Free to Play: players can play and earn cards with no purchase necessary. On top of that, Parallel compensates players that play for free by automatically entering them into weekly and monthly giveaways every season. This is on top of rewards players can get through in-game tournaments, battle pass, referring friends and winning ranked matches. Bank Cards for Energy: players strategically choose which cards to bank each turn, gaining energy that allows them to play units, relics, effects and upgrades. Players will find that choosing what card to bank and what card to play is critical, as the right combination can create a powerful winning strategy. Paragons & Factions: Parallel has 5 unique factions with 15 game changing leaders (Paragons) each with their own active/passive abilities helping players craft a unique play style. Play At Your Pace: Parallel has different game modes for all types of players. Beginners and casual players can enjoy modes like unranked, AI and rookie. Competitive players can participate in the Ranked Ladder and Tournament Modes to earn exclusive rewards and prizes. Free to Play: players can play and earn cards with no purchase necessary. On top of that, Parallel compensates players that play for free by automatically entering them into weekly and monthly giveaways every season. This is on top of rewards players can get through in-game tournaments, battle pass, referring friends and winning ranked matches. Bank Cards for Energy: players strategically choose which cards to bank each turn, gaining energy that allows them to play units, relics, effects and upgrades. Players will find that choosing what card to bank and what card to play is critical, as the right combination can create a powerful winning strategy. Paragons & Factions: Parallel has 5 unique factions with 15 game changing leaders (Paragons) each with their own active/passive abilities helping players craft a unique play style. Play At Your Pace: Parallel has different game modes for all types of players. Beginners and casual players can enjoy modes like unranked, AI and rookie. Competitive players can participate in the Ranked Ladder and Tournament Modes to earn exclusive rewards and prizes. Parallel is truly built for TCG veterans and new players to the genre. Founded in 2021, Parallel has already seen quite a bit of success. Despite its TCG being in open beta, the game has already won several game of the year awards. The studio has also announced a prize pool of $500,000 for its 2024 competitive scene. Along with that, several professional players and prominent creators that played Hearthstone have joined Parallel’s ambassador program, priming the game to be the next big TCG. Finally, the studio recently released an expansion set called "Planetfall" that is currently playable. The expansion set added over 100 new cards to the game and the studio plans to launch more expansion sets in the future. Quotes From Parallel "Our journey has been nothing short of extraordinary, and being on Epic marks a significant milestone in our global expansion strategy," says Sascha M, CEO & Co-Founder of Parallel. "Our journey has been nothing short of extraordinary, and being on Epic marks a significant milestone in our global expansion strategy," says Sascha M, CEO & Co-Founder of Parallel. “Parallel going live on Epic opens us up to millions more players worldwide. This expansion into Epic comes alongside the kickoff of our global competitive series, the Parallel League, which has over $500,000 in prizes in its first year. But we don't plan to stop there. As we roll out more ways to access and play Parallel we also plan to scale our competitive prize pool into millions of dollars next year.” - Franklin Fitch, Head of Partnerships, Parallel Studios “Parallel going live on Epic opens us up to millions more players worldwide. This expansion into Epic comes alongside the kickoff of our global competitive series, the Parallel League, which has over $500,000 in prizes in its first year. But we don't plan to stop there. As we roll out more ways to access and play Parallel we also plan to scale our competitive prize pool into millions of dollars next year.” - Franklin Fitch, Head of Partnerships, Parallel Studios About Parallel Studios Parallel Studios is a revolutionary game development company creating the next AAA sci-fi franchise. Current projects include the Parallel TCG, an exciting science fiction card game in which five factions battle for supremacy over a future Earth, and Parallel Colony, an AI-powered idle game in which you play through conversation with your Avatar. Parallel aims to continue breaking new ground and bringing amazing experiences to this new IP. Contact Head of Growth Swish Goswami Parallel Studios swish@parallel.game 3472579158 This story was distributed as a release by Gamingwire under HackerNoon’s Business Blogging Program. Learn more about the program here. This story was distributed as a release by Gamingwire under HackerNoon’s Business Blogging Program. Learn more about the program here . here here

Parallel TCG Releases On Epic Games Store

Authors:
(1) Simon Kafader, University of Bern, Bern, Switzerland (simon.kafader@inf.unibe.ch);
(2) Mohammad Ghafari, University of Auckland, Auckland, New Zealand (m.ghafari@auckland.ac.nz). Table of Links Abstract and I. Introduction
II. Related Work
III. Background and Motivation
IV. FluentCrypto
V. Study Setup
VI. Study Result
VII. Threats to Validity
VIII. Conclusion and References II. RELATED WORK We discuss related work from two perspectives. Firstly, we discuss an excerpt of literature that investigated why developers struggle with using crypto APIs. Secondly, we discuss previous work that aimed to help developers in circumventing crypto misuses. Nevertheless, to the best of our knowledge, there is no such research on Node.js cryptography. A. API design Nadi et al. conducted a survey with 11 developers who posted crypto-related questions on the Stack Overflow website, and 37 developers who had experience with Java’s cryptography APIs [15]. They realized that developers are able to understand crypto concepts but they fail to adopt them in practice. They concluded that the concepts behind crypto APIs are low-level and developers welcome task-based solutions for working with crypto APIs. Green and Smith found that many misuses of cryptographic libraries originate in the developer having trouble understanding the API [16]. They criticized that while it has become accepted that systems should be user-friendly for the end user, a different attitude where the end-user is expected to be an expert prevails amongst cryptographic libraries. They then proposed a set of characteristics that should lead to more secure cryptographic APIs, including: the API should be easy to use, even without documentation; incorrect use should lead to visible errors; defaults should be safe and never ambiguous; and code that uses the API should be easy to read and update. Das et al. selected the most popular cryptographic libraries from C, C++, Java, Python and Go and examined them for properties that cause cryptographic misuse by developers [17]. They examined the libraries based on a set of common potential issues such as initialization vector reuse, library defaults or incomplete features. Their work highlighted the disconnect between the actual user of such a library and the user for whom it is designed. B. Developer support Several tools have been developed to support developers in using cryptography. For example, Krüger et al. developed the CogniCrypt Eclipse plug-in that generates code snippets for cryptography tasks [18]. Nguyen et al developed FixDroid which helps developers in fixing crypto-related issues in the AndroidStudio IDE [19]. Hazhirpasand et al. developed an interactive web platform named CryptoExplorer that provides developers with real-world examples, specifically 3263 secure and 5897 insecure uses of Java Cryptography Architecture [14]. Singleton et al. developed CryptoTutor, an educational tool that flags common cryptographic misuses and suggests possible repairs [20]. The authors discussed how such a tool can be integrated into programming courses to improve developer knowledge in the cryptography domain. Several tools also exist to detect crypto misuses. For instance, Rahaman et al. developed CryptoGuard to identify crypto issues statically [9]. Piccolboni et al. proposed CryLogger, an open-source tool to detect crypto misuses dynamically [21]. Gorski surveyed 25 professional software developers to identify what kind of feedback is helpful to them in avoiding crypto misuses during programming [22]. They found that participants appreciate a clear warning message (including title, content, and code location) that is tailored to the context. This paper is available on arxiv under CC BY 4.0 DEED license. Authors: (1) Simon Kafader, University of Bern, Bern, Switzerland (simon.kafader@inf.unibe.ch); (2) Mohammad Ghafari, University of Auckland, Auckland, New Zealand (m.ghafari@auckland.ac.nz). Authors: Authors: (1) Simon Kafader, University of Bern, Bern, Switzerland (simon.kafader@inf.unibe.ch); (2) Mohammad Ghafari, University of Auckland, Auckland, New Zealand (m.ghafari@auckland.ac.nz). Table of Links Abstract and I. Introduction II. Related Work III. Background and Motivation IV. FluentCrypto V. Study Setup VI. Study Result VII. Threats to Validity VIII. Conclusion and References Abstract and I. Introduction Abstract and I. Introduction II. Related Work II. Related Work III. Background and Motivation III. Background and Motivation IV. FluentCrypto IV. FluentCrypto V. Study Setup V. Study Setup VI. Study Result VI. Study Result VII. Threats to Validity VII. Threats to Validity VIII. Conclusion and References VIII. Conclusion and References II. RELATED WORK We discuss related work from two perspectives. Firstly, we discuss an excerpt of literature that investigated why developers struggle with using crypto APIs. Secondly, we discuss previous work that aimed to help developers in circumventing crypto misuses. Nevertheless, to the best of our knowledge, there is no such research on Node.js cryptography. A. API design A. API design Nadi et al. conducted a survey with 11 developers who posted crypto-related questions on the Stack Overflow website, and 37 developers who had experience with Java’s cryptography APIs [15]. They realized that developers are able to understand crypto concepts but they fail to adopt them in practice. They concluded that the concepts behind crypto APIs are low-level and developers welcome task-based solutions for working with crypto APIs. Green and Smith found that many misuses of cryptographic libraries originate in the developer having trouble understanding the API [16]. They criticized that while it has become accepted that systems should be user-friendly for the end user, a different attitude where the end-user is expected to be an expert prevails amongst cryptographic libraries. They then proposed a set of characteristics that should lead to more secure cryptographic APIs, including: the API should be easy to use, even without documentation; incorrect use should lead to visible errors; defaults should be safe and never ambiguous; and code that uses the API should be easy to read and update. Das et al. selected the most popular cryptographic libraries from C, C++, Java, Python and Go and examined them for properties that cause cryptographic misuse by developers [17]. They examined the libraries based on a set of common potential issues such as initialization vector reuse, library defaults or incomplete features. Their work highlighted the disconnect between the actual user of such a library and the user for whom it is designed. B. Developer support B. Developer support Several tools have been developed to support developers in using cryptography. For example, Krüger et al. developed the CogniCrypt Eclipse plug-in that generates code snippets for cryptography tasks [18]. Nguyen et al developed FixDroid which helps developers in fixing crypto-related issues in the AndroidStudio IDE [19]. Hazhirpasand et al. developed an interactive web platform named CryptoExplorer that provides developers with real-world examples, specifically 3263 secure and 5897 insecure uses of Java Cryptography Architecture [14]. Singleton et al. developed CryptoTutor, an educational tool that flags common cryptographic misuses and suggests possible repairs [20]. The authors discussed how such a tool can be integrated into programming courses to improve developer knowledge in the cryptography domain. Several tools also exist to detect crypto misuses. For instance, Rahaman et al. developed CryptoGuard to identify crypto issues statically [9]. Piccolboni et al. proposed CryLogger, an open-source tool to detect crypto misuses dynamically [21]. Gorski surveyed 25 professional software developers to identify what kind of feedback is helpful to them in avoiding crypto misuses during programming [22]. They found that participants appreciate a clear warning message (including title, content, and code location) that is tailored to the context. This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv

FluentCrypto: Cryptography in Easy Mode: Related Work

Authors:
(1) Shadab Ahamed, University of British Columbia, Vancouver, BC, Canada, BC Cancer Research Institute, Vancouver, BC, Canada. He was also a Mitacs Accelerate Fellow (May 2022 - April 2023) with Microsoft AI for Good Lab, Redmond, WA, USA (e-mail: shadabahamed1996@gmail.com);
(2) Yixi Xu, Microsoft AI for Good Lab, Redmond, WA, USA;
(3) Claire Gowdy, BC Children’s Hospital, Vancouver, BC, Canada;
(4) Joo H. O, St. Mary’s Hospital, Seoul, Republic of Korea;
(5) Ingrid Bloise, BC Cancer, Vancouver, BC, Canada;
(6) Don Wilson, BC Cancer, Vancouver, BC, Canada;
(7) Patrick Martineau, BC Cancer, Vancouver, BC, Canada;
(8) Franc¸ois Benard, BC Cancer, Vancouver, BC, Canada;
(9) Fereshteh Yousefirizi, BC Cancer Research Institute, Vancouver, BC, Canada;
(10) Rahul Dodhia, Microsoft AI for Good Lab, Redmond, WA, USA;
(11) Juan M. Lavista, Microsoft AI for Good Lab, Redmond, WA, USA;
(12) William B. Weeks, Microsoft AI for Good Lab, Redmond, WA, USA;
(13) Carlos F. Uribe, BC Cancer Research Institute, Vancouver, BC, Canada, and University of British Columbia, Vancouver, BC, Canada;
(14) Arman Rahmim, BC Cancer Research Institute, Vancouver, BC, Canada, and University of British Columbia, Vancouver, BC, Canada. Table of Links Abstract and Intro
Related Work
Materials and Methods
Results
Discussion
Conclusion and References II. RELATED WORK Numerous works have explored the application of deep learning methods for segmenting lymphoma in PET/CT images. Yuan et al. [4] developed a feature fusion technique to utilize the complementary information from multi-modality data. Hu et al. [5] proposed fusing a combination of 3D ResUNet trained on volumetric data and three 2D ResUNet trained on 2D slices from three orthogonal directions for enhancing segmentation performance. Li et al. [6] proposed DenseX-Net trained in an end-to-end fashion integrating supervised and unsupervised methods for lymphoma detection and segmentation. Liu et al. [7] introduced techniques such as patch-based negative sample augmentation and label guidance for training a 3D Residual-UNet for lymphoma segmentation. A major limitation of all these works was that they were developed on relatively smaller-sized datasets (less than 100 images). Moreover, most of these methods did not compare the performance of their proposed methods with other baselines or with the performance of physicians. Constantino et al. [8] compared the performances of 7 semi-automated and 2 deep learning segmentation methods, while Weisman et al. [9] compared 11 automated segmentation techniques, although both these studies were performed on smaller datasets of sizes 65 and 90 respectively. Weisman et al. [10] compared the segmentation performances of automated 3D Deep Medic method with that of physician although even this study included just 90 lymphoma cases. Except for [10], none of these studies reported model generalization on out-ofdistribution dataset (such as on data collected from different centers), limiting their robustness quantification and external validity. Jiang et al. [11] used a relatively larger dataset as compared to the above studies with 297 images to train a 3D UNet. They even performed out-of-distribution testing on 117 images collected from a different center. To the best of our knowledge, the largest lymphoma PET/CT dataset for deep learning-based lesion segmentation ever reported is the work by Blanc-Durand et al. [12] who used 639 images for model development and 94 for external testing; however, this study only used standard segmentation evaluation metrics and assessed their model’s ability for predicting accurate TMTV. Both the studies [11] and [12] are limited by the fact that their datasets exclusively consisted of patients diagnosed with diffuse large B-cell lymphoma (DLBCL), representing only a single subtype of lymphoma. Most of the existing studies on deep learning-based lymphoma segmentation report their performances on generic segmentation metrics such as Dice similarity coefficient (DSC), intersection-over-union (IoU), sensitivity, etc. In the presence of large segmented lesions, very small missed lesions or small false positives do not contribute much to the DSC value. Hence, there is a need to report the volumes of false positives and false negatives. It will also be beneficial to evaluate the detection performances on a per-lesion basis (number of connected components detected vs missed), since automated detection of even a few voxels of all lesions can help physicians quickly locate the regions of interest, even if the DSC is low. Moreover, the difficulty of the segmentation/detection task is often not assessed via inter- or intra-observer agreement analysis. Our study aims to address these limitations. We trained and validated four deep neural networks on lymphoma PET/CT datasets from three cohorts, encompassing two distinct subtypes of lymphoma: DLBCL and primary mediastinal large B-cell lymphoma (PMBCL). (i) We performed both in (images coming from same cohorts as the training/validation set) and out-of-distribution or external (images from a fourth cohort not used for training/validation) testing to evaluate the robustness of our models. (ii) We reported the performance using DSC, volumes of false positives and negatives, and evaluated the performance dependence on six different types of lesion measures. (iii) We also evaluated the ability of our networks to reproduce these ground truth lesion measures and computed networks’ error in predicting them. (iv) We proposed three types of detection criteria for our use-case and evaluate model’s performance on these metrics. (v) Finally, we evaluated the intra- and inter-observer agreement to give a measure of the difficulty of lesion segmentation task on our datasets. This paper is available on arxiv under CC 4.0 license. Authors: (1) Shadab Ahamed, University of British Columbia, Vancouver, BC, Canada, BC Cancer Research Institute, Vancouver, BC, Canada. He was also a Mitacs Accelerate Fellow (May 2022 - April 2023) with Microsoft AI for Good Lab, Redmond, WA, USA (e-mail: shadabahamed1996@gmail.com); (2) Yixi Xu, Microsoft AI for Good Lab, Redmond, WA, USA; (3) Claire Gowdy, BC Children’s Hospital, Vancouver, BC, Canada; (4) Joo H. O, St. Mary’s Hospital, Seoul, Republic of Korea; (5) Ingrid Bloise, BC Cancer, Vancouver, BC, Canada; (6) Don Wilson, BC Cancer, Vancouver, BC, Canada; (7) Patrick Martineau, BC Cancer, Vancouver, BC, Canada; (8) Franc¸ois Benard, BC Cancer, Vancouver, BC, Canada; (9) Fereshteh Yousefirizi, BC Cancer Research Institute, Vancouver, BC, Canada; (10) Rahul Dodhia, Microsoft AI for Good Lab, Redmond, WA, USA; (11) Juan M. Lavista, Microsoft AI for Good Lab, Redmond, WA, USA; (12) William B. Weeks, Microsoft AI for Good Lab, Redmond, WA, USA; (13) Carlos F. Uribe, BC Cancer Research Institute, Vancouver, BC, Canada, and University of British Columbia, Vancouver, BC, Canada; (14) Arman Rahmim, BC Cancer Research Institute, Vancouver, BC, Canada, and University of British Columbia, Vancouver, BC, Canada. Authors: Authors: (1) Shadab Ahamed, University of British Columbia, Vancouver, BC, Canada, BC Cancer Research Institute, Vancouver, BC, Canada. He was also a Mitacs Accelerate Fellow (May 2022 - April 2023) with Microsoft AI for Good Lab, Redmond, WA, USA (e-mail: shadabahamed1996@gmail.com); (2) Yixi Xu, Microsoft AI for Good Lab, Redmond, WA, USA; (3) Claire Gowdy, BC Children’s Hospital, Vancouver, BC, Canada; (4) Joo H. O, St. Mary’s Hospital, Seoul, Republic of Korea; (5) Ingrid Bloise, BC Cancer, Vancouver, BC, Canada; (6) Don Wilson, BC Cancer, Vancouver, BC, Canada; (7) Patrick Martineau, BC Cancer, Vancouver, BC, Canada; (8) Franc¸ois Benard, BC Cancer, Vancouver, BC, Canada; (9) Fereshteh Yousefirizi, BC Cancer Research Institute, Vancouver, BC, Canada; (10) Rahul Dodhia, Microsoft AI for Good Lab, Redmond, WA, USA; (11) Juan M. Lavista, Microsoft AI for Good Lab, Redmond, WA, USA; (12) William B. Weeks, Microsoft AI for Good Lab, Redmond, WA, USA; (13) Carlos F. Uribe, BC Cancer Research Institute, Vancouver, BC, Canada, and University of British Columbia, Vancouver, BC, Canada; (14) Arman Rahmim, BC Cancer Research Institute, Vancouver, BC, Canada, and University of British Columbia, Vancouver, BC, Canada. Table of Links Abstract and Intro Related Work Materials and Methods Results Discussion Conclusion and References Abstract and Intro Abstract and Intro Related Work Related Work Materials and Methods Materials and Methods Results Results Discussion Discussion Conclusion and References Conclusion and References II. RELATED WORK Numerous works have explored the application of deep learning methods for segmenting lymphoma in PET/CT images. Yuan et al. [4] developed a feature fusion technique to utilize the complementary information from multi-modality data. Hu et al. [5] proposed fusing a combination of 3D ResUNet trained on volumetric data and three 2D ResUNet trained on 2D slices from three orthogonal directions for enhancing segmentation performance. Li et al. [6] proposed DenseX-Net trained in an end-to-end fashion integrating supervised and unsupervised methods for lymphoma detection and segmentation. Liu et al. [7] introduced techniques such as patch-based negative sample augmentation and label guidance for training a 3D Residual-UNet for lymphoma segmentation. A major limitation of all these works was that they were developed on relatively smaller-sized datasets (less than 100 images). Moreover, most of these methods did not compare the performance of their proposed methods with other baselines or with the performance of physicians. Constantino et al. [8] compared the performances of 7 semi-automated and 2 deep learning segmentation methods, while Weisman et al. [9] compared 11 automated segmentation techniques, although both these studies were performed on smaller datasets of sizes 65 and 90 respectively. Weisman et al. [10] compared the segmentation performances of automated 3D Deep Medic method with that of physician although even this study included just 90 lymphoma cases. Except for [10], none of these studies reported model generalization on out-ofdistribution dataset (such as on data collected from different centers), limiting their robustness quantification and external validity. Jiang et al. [11] used a relatively larger dataset as compared to the above studies with 297 images to train a 3D UNet. They even performed out-of-distribution testing on 117 images collected from a different center. To the best of our knowledge, the largest lymphoma PET/CT dataset for deep learning-based lesion segmentation ever reported is the work by Blanc-Durand et al. [12] who used 639 images for model development and 94 for external testing; however, this study only used standard segmentation evaluation metrics and assessed their model’s ability for predicting accurate TMTV. Both the studies [11] and [12] are limited by the fact that their datasets exclusively consisted of patients diagnosed with diffuse large B-cell lymphoma (DLBCL), representing only a single subtype of lymphoma. Most of the existing studies on deep learning-based lymphoma segmentation report their performances on generic segmentation metrics such as Dice similarity coefficient (DSC), intersection-over-union (IoU), sensitivity, etc. In the presence of large segmented lesions, very small missed lesions or small false positives do not contribute much to the DSC value. Hence, there is a need to report the volumes of false positives and false negatives. It will also be beneficial to evaluate the detection performances on a per-lesion basis (number of connected components detected vs missed), since automated detection of even a few voxels of all lesions can help physicians quickly locate the regions of interest, even if the DSC is low. Moreover, the difficulty of the segmentation/detection task is often not assessed via inter- or intra-observer agreement analysis. Our study aims to address these limitations. We trained and validated four deep neural networks on lymphoma PET/CT datasets from three cohorts, encompassing two distinct subtypes of lymphoma: DLBCL and primary mediastinal large B-cell lymphoma (PMBCL). (i) We performed both in (images coming from same cohorts as the training/validation set) and out-of-distribution or external (images from a fourth cohort not used for training/validation) testing to evaluate the robustness of our models. (ii) We reported the performance using DSC, volumes of false positives and negatives, and evaluated the performance dependence on six different types of lesion measures. (iii) We also evaluated the ability of our networks to reproduce these ground truth lesion measures and computed networks’ error in predicting them. (iv) We proposed three types of detection criteria for our use-case and evaluate model’s performance on these metrics. (v) Finally, we evaluated the intra- and inter-observer agreement to give a measure of the difficulty of lesion segmentation task on our datasets. This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Deep Neural Networks to Detect and Quantify Lymphoma Lesions: Related Work

ডিপ নিউরাল নেটওয়ার্ক লিম্ফোমা ক্ষত সনাক্ত এবং পরিমাপ করতে: সম্পর্কিত কাজ

Tiefe neuronale Netzwerke zur Erkennung und Quantifizierung von Lymphomläsionen: Verwandte Arbeiten

Redes neuronales profundas para detectar y cuantificar lesiones de linfoma: trabajo relacionado

Réseaux de neurones profonds pour détecter et quantifier les lésions de lymphome : travaux connexes

लिम्फोमा घावों का पता लगाने और उनकी मात्रा निर्धारित करने के लिए डीप न्यूरल नेटवर्क: संबंधित कार्य

リンパ腫病変の検出と定量化のためのディープニューラルネットワーク: 関連研究

림프종 병변을 탐지하고 정량화하기 위한 심층 신경망: 관련 연구

Redes Neurais Profundas para Detectar e Quantificar Lesões de Linfoma: Trabalho Relacionado

Глубокие нейронные сети для обнаружения и количественной оценки поражений лимфомой: соответствующая работа

Lenfoma Lezyonlarını Tespit ve Ölçmek İçin Derin Sinir Ağları: İlgili Çalışma

Mạng lưới thần kinh sâu để phát hiện và định lượng tổn thương ung thư hạch: Công việc liên quan

深度神经网络检测和量化淋巴瘤病变：相关工作

Author:
(1) Katrin Fischer, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: katrinfi@usc.edu);
(2) Donggyu Kim, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: donggyuk@usc.edu);
(3) Joo-Wha Hong, Marshall School of Business at the University of Southern California, Los Angeles (Email: joowhaho@marshall.usc.edu). Table of Links Abstract Introduction & Related Work Method Analysis & Results Discussion & Conclusion References Abstract— As social and socially assistive robots are becoming more prevalent in our society, it is beneficial to understand how people form first impressions of them and eventually come to trust and accept them. This paper describes an Amazon Mechanical Turk study (n = 239) that investigated trust and its antecedents trustworthiness and first impressions. Participants evaluated the social robot Pepper’s warmth and competence as well as trustworthiness characteristics ability, benevolence and integrity followed by their trust in and intention to use the robot. Mediation analyses assessed to what degree participants’ first impressions affected their willingness to trust and use it. Known constructs from user acceptance and trust research were introduced to explain the pathways in which one perception predicted the next. Results showed that trustworthiness and trust, in serial, mediated the relationship between first impressions and behavioral intention. I. INTRODUCTION & RELATED WORK Trust plays an important role in human relationships and is similarly important in establishing relationships between humans and robots. It is “the confidence that one will find what is desired from another, rather than what is feared” [1] and is crucial in the presence of (perceived) risk, e.g. when another’s abilities or actions cannot be foreseen [2]. Consistent and predictable behaviors as well as trust-building characteristics such as dependability are cornerstones that make up trust in human relationships [3]. Research on human-robot trust has generated a sizeable body of literature in which multiple definitions and frameworks of trust exist. For instance, Lee & See applied Rempel et al.’s [3] dimensions of human trust to interactions with robots in terms of their performance, behavior determining operations (algorithms) and the degree to which they are used within their designer’s intent [4]. Other research has considered the multidimensional nature of trust with a focus on gains and losses in human-robot trust due to robot behaviour change [5]. We follow the model of Mayer et al. [6] who argued that trust is built through evaluating trustworthiness characteristics of the trustee, which has been shown to apply to HRI contexts by evaluating the robot’s ability, integrity and benevolence [7]. Trustworthiness is also connected to social dimensions of human perceptions of robots such as warmth and competence which vary based on robot appearance [8]. According to the stereotype content model [9], these first impressions help us determine whether a new acquaintance is likely to constitute a friend or a foe. Warmth judgements (trustworthiness, helpfulness, perceived intent) occur first and decide affective and behavioral reactions, whereas perceived competence ascertains to what extent the other can act on their motives (perceived ability, efficiency, intelligence). Competence and warmth stereotypes predict emotions, which directly predict behaviors [10] and they apply to people as well as to social robots [11]. There is a relationship between warmth and trust in that somebody that is perceived to be warm is simultaneously considered trustworthy, or its inverse, cold and untrustworthy [9]. Trust plays a crucial role in HRI [12], [13] and has been recognized as a factor that not only predicts the quality of the interaction, but also how willing people are to use social robots for certain tasks [14]–[16]. Among the most commonly employed models to assess use and acceptance of new technologies are TAM (Technology Acceptance Model) [17] and UTAUT (Unified Theory of Acceptance and Use of Technology) [18]. TAM studies can explain approximately 50% of the variance in technology acceptance outcomes while UTAUT has been found to produce an adjusted R2 of over 69% [18]. Its outcome, intention to use, is relevant to HRI research as a predictor of robot acceptance, on which, together with trust, the success of socially assistive robots depends [19]. Past research has looked at trust and acceptance in HRI, but not explored the specific pathways of their relationship. RQ1. How do the above identified constructs of trust, trustworthiness, warmth, competence, and intention to use social robots interrelate? RQ2. What are the antecedents of trust? Of social robot acceptance? This paper is available on arxiv under CC 4.0 license. Author: (1) Katrin Fischer, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: katrinfi@usc.edu); (2) Donggyu Kim, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: donggyuk@usc.edu); (3) Joo-Wha Hong, Marshall School of Business at the University of Southern California, Los Angeles (Email: joowhaho@marshall.usc.edu). Author: Author: (1) Katrin Fischer, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: katrinfi@usc.edu); (2) Donggyu Kim, Annenberg School for Communication at the University of Southern California, Los Angeles (Email: donggyuk@usc.edu); (3) Joo-Wha Hong, Marshall School of Business at the University of Southern California, Los Angeles (Email: joowhaho@marshall.usc.edu). Table of Links Abstract Introduction & Related Work Abstract Introduction & Related Work Method Method Analysis & Results Analysis & Results Discussion & Conclusion Discussion & Conclusion References References Abstract— As social and socially assistive robots are becoming more prevalent in our society, it is beneficial to understand how people form first impressions of them and eventually come to trust and accept them. This paper describes an Amazon Mechanical Turk study (n = 239) that investigated trust and its antecedents trustworthiness and first impressions. Participants evaluated the social robot Pepper’s warmth and competence as well as trustworthiness characteristics ability, benevolence and integrity followed by their trust in and intention to use the robot. Mediation analyses assessed to what degree participants’ first impressions affected their willingness to trust and use it. Known constructs from user acceptance and trust research were introduced to explain the pathways in which one perception predicted the next. Results showed that trustworthiness and trust, in serial, mediated the relationship between first impressions and behavioral intention. Abstract — As social and socially assistive robots are becoming more prevalent in our society, it is beneficial to understand how people form first impressions of them and eventually come to trust and accept them. This paper describes an Amazon Mechanical Turk study (n = 239) that investigated trust and its antecedents trustworthiness and first impressions. Participants evaluated the social robot Pepper’s warmth and competence as well as trustworthiness characteristics ability, benevolence and integrity followed by their trust in and intention to use the robot. Mediation analyses assessed to what degree participants’ first impressions affected their willingness to trust and use it. Known constructs from user acceptance and trust research were introduced to explain the pathways in which one perception predicted the next. Results showed that trustworthiness and trust, in serial, mediated the relationship between first impressions and behavioral intention. Abstract I. INTRODUCTION & RELATED WORK Trust plays an important role in human relationships and is similarly important in establishing relationships between humans and robots. It is “the confidence that one will find what is desired from another, rather than what is feared” [1] and is crucial in the presence of (perceived) risk, e.g. when another’s abilities or actions cannot be foreseen [2]. Consistent and predictable behaviors as well as trust-building characteristics such as dependability are cornerstones that make up trust in human relationships [3]. Research on human-robot trust has generated a sizeable body of literature in which multiple definitions and frameworks of trust exist. For instance, Lee & See applied Rempel et al.’s [3] dimensions of human trust to interactions with robots in terms of their performance, behavior determining operations (algorithms) and the degree to which they are used within their designer’s intent [4]. Other research has considered the multidimensional nature of trust with a focus on gains and losses in human-robot trust due to robot behaviour change [5]. We follow the model of Mayer et al. [6] who argued that trust is built through evaluating trustworthiness characteristics of the trustee, which has been shown to apply to HRI contexts by evaluating the robot’s ability, integrity and benevolence [7]. Trustworthiness is also connected to social dimensions of human perceptions of robots such as warmth and competence which vary based on robot appearance [8]. According to the stereotype content model [9], these first impressions help us determine whether a new acquaintance is likely to constitute a friend or a foe. Warmth judgements (trustworthiness, helpfulness, perceived intent) occur first and decide affective and behavioral reactions, whereas perceived competence ascertains to what extent the other can act on their motives (perceived ability, efficiency, intelligence). Competence and warmth stereotypes predict emotions, which directly predict behaviors [10] and they apply to people as well as to social robots [11]. There is a relationship between warmth and trust in that somebody that is perceived to be warm is simultaneously considered trustworthy, or its inverse, cold and untrustworthy [9]. Trust plays a crucial role in HRI [12], [13] and has been recognized as a factor that not only predicts the quality of the interaction, but also how willing people are to use social robots for certain tasks [14]–[16]. Among the most commonly employed models to assess use and acceptance of new technologies are TAM (Technology Acceptance Model) [17] and UTAUT (Unified Theory of Acceptance and Use of Technology) [18]. TAM studies can explain approximately 50% of the variance in technology acceptance outcomes while UTAUT has been found to produce an adjusted R2 of over 69% [18]. Its outcome, intention to use, is relevant to HRI research as a predictor of robot acceptance, on which, together with trust, the success of socially assistive robots depends [19]. Past research has looked at trust and acceptance in HRI, but not explored the specific pathways of their relationship. RQ1. How do the above identified constructs of trust, trustworthiness, warmth, competence, and intention to use social robots interrelate? RQ1. RQ2. What are the antecedents of trust? Of social robot acceptance? RQ2. This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Trust and Acceptance of Social Robots: Abstract, Introduction & Related Work

Authors:
(1) TIMNIT GEBRU, Black in AI;
(2) JAMIE MORGENSTERN, University of Washington;
(3) BRIANA VECCHIONE, Cornell University;
(4) JENNIFER WORTMAN VAUGHAN, Microsoft Research;
(5) HANNA WALLACH, Microsoft Research;
(6) HAL DAUMÉ III, Microsoft Research; University of Maryland;
(7) KATE CRAWFORD, Microsoft Research. Table of Links 1 Introduction 1.1 Objectives 2 Development Process 3 Questions and Workflow 3.1 Motivation 3.2 Composition 3.3 Collection Process 3.4 Preprocessing/cleaning/labeling 3.5 Uses 3.6 Distribution 3.7 Maintenance 4 Impact and Challenges Acknowledgments and References Appendix 3.2 Composition Dataset creators should read through these questions prior to any data collection and then provide answers once data collection is complete. Most of the questions in this section are intended to provide dataset consumers with the information they need to make informed decisions about using the dataset for their chosen tasks. Some of the questions are designed to elicit information about compliance with the EU’s General Data Protection Regulation (GDPR) or comparable regulations in other jurisdictions. Questions that apply only to datasets that relate to people are grouped together at the end of the section. We recommend taking a broad interpretation of whether a dataset relates to people. For example, any dataset containing text that was written by people relates to people. • What do the instances that comprise the dataset represent (e.g., documents, photos, people, countries)? Are there multiple types of instances (e.g., movies, users, and ratings; people and interactions between them; nodes and edges)? Please provide a description. • How many instances are there in total (of each type, if appropriate)? • Does the dataset contain all possible instances or is it a sample (not necessarily random) of instances from a larger set? If the dataset is a sample, then what is the larger set? Is the sample representative of the larger set (e.g., geographic coverage)? If so, please describe how this representativeness was validated/verified. If it is not representative of the larger set, please describe why not (e.g., to cover a more diverse range of instances, because instances were withheld or unavailable). • What data does each instance consist of? “Raw” data (e.g., unprocessed text or images) or features? In either case, please provide a description. • Is there a label or target associated with each instance? If so, please provide a description. • Is any information missing from individual instances? If so, please provide a description, explaining why this information is missing (e.g., because it was unavailable). This does not include intentionally removed information, but might include, e.g., redacted text. • Are relationships between individual instances made explicit (e.g., users’ movie ratings, social network links)? If so, please describe how these relationships are made explicit. • Are there recommended data splits (e.g., training, development/validation, testing)? If so, please provide a description of these splits, explaining the rationale behind them. • Are there any errors, sources of noise, or redundancies in the dataset? If so, please provide a description. • Is the dataset self-contained, or does it link to or otherwise rely on external resources (e.g., websites, tweets, other datasets)? If it links to or relies on external resources, a) are there guarantees that they will exist, and remain constant, over time; b) are there official archival versions of the complete dataset (i.e., including the external resources as they existed at the time the dataset was created); c) are there any restrictions (e.g., licenses, fees) associated with any of the external resources that might apply to a dataset consumer? Please provide descriptions of all external resources and any restrictions associated with them, as well as links or other access points, as appropriate. • Does the dataset contain data that might be considered confidential (e.g., data that is protected by legal privilege or by doctor– patient confidentiality, data that includes the content of individuals’ non-public communications)? If so, please provide a description. • Does the dataset contain data that, if viewed directly, might be offensive, insulting, threatening, or might otherwise cause anxiety? If so, please describe why. If the dataset does not relate to people, you may skip the remaining questions in this section. • Does the dataset identify any subpopulations (e.g., by age, gender)? If so, please describe how these subpopulations are identified and provide a description of their respective distributions within the dataset. • Is it possible to identify individuals (i.e., one or more natural persons), either directly or indirectly (i.e., in combination with other data) from the dataset? If so, please describe how. • Does the dataset contain data that might be considered sensitive in any way (e.g., data that reveals race or ethnic origins, sexual orientations, religious beliefs, political opinions or union memberships, or locations; financial or health data; biometric or genetic data; forms of government identification, such as social security numbers; criminal history)? If so, please provide a description. • Any other comments? This paper is available on arxiv under CC 4.0 license. Authors: (1) TIMNIT GEBRU, Black in AI; (2) JAMIE MORGENSTERN, University of Washington; (3) BRIANA VECCHIONE, Cornell University; (4) JENNIFER WORTMAN VAUGHAN, Microsoft Research; (5) HANNA WALLACH, Microsoft Research; (6) HAL DAUMÉ III, Microsoft Research; University of Maryland; (7) KATE CRAWFORD, Microsoft Research. Authors: Authors: (1) TIMNIT GEBRU, Black in AI; (2) JAMIE MORGENSTERN, University of Washington; (3) BRIANA VECCHIONE, Cornell University; (4) JENNIFER WORTMAN VAUGHAN, Microsoft Research; (5) HANNA WALLACH, Microsoft Research; (6) HAL DAUMÉ III, Microsoft Research; University of Maryland; (7) KATE CRAWFORD, Microsoft Research. Table of Links 1 Introduction 1 Introduction 1.1 Objectives 1.1 Objectives 2 Development Process 2 Development Process 3 Questions and Workflow 3 Questions and Workflow 3.1 Motivation 3.1 Motivation 3.2 Composition 3.2 Composition 3.3 Collection Process 3.3 Collection Process 3.4 Preprocessing/cleaning/labeling 3.4 Preprocessing/cleaning/labeling 3.5 Uses 3.5 Uses 3.6 Distribution 3.6 Distribution 3.7 Maintenance 3.7 Maintenance 4 Impact and Challenges 4 Impact and Challenges Acknowledgments and References Acknowledgments and References Appendix Appendix 3.2 Composition Dataset creators should read through these questions prior to any data collection and then provide answers once data collection is complete. Most of the questions in this section are intended to provide dataset consumers with the information they need to make informed decisions about using the dataset for their chosen tasks. Some of the questions are designed to elicit information about compliance with the EU’s General Data Protection Regulation (GDPR) or comparable regulations in other jurisdictions. Questions that apply only to datasets that relate to people are grouped together at the end of the section. We recommend taking a broad interpretation of whether a dataset relates to people. For example, any dataset containing text that was written by people relates to people. • What do the instances that comprise the dataset represent (e.g., documents, photos, people, countries)? Are there multiple types of instances (e.g., movies, users, and ratings; people and interactions between them; nodes and edges)? Please provide a description. • What do the instances that comprise the dataset represent (e.g., documents, photos, people, countries)? • How many instances are there in total (of each type, if appropriate)? • How many instances are there in total (of each type, if appropriate)? • Does the dataset contain all possible instances or is it a sample (not necessarily random) of instances from a larger set? If the dataset is a sample, then what is the larger set? Is the sample representative of the larger set (e.g., geographic coverage)? If so, please describe how this representativeness was validated/verified. If it is not representative of the larger set, please describe why not (e.g., to cover a more diverse range of instances, because instances were withheld or unavailable). • Does the dataset contain all possible instances or is it a sample (not necessarily random) of instances from a larger set? • What data does each instance consist of? “Raw” data (e.g., unprocessed text or images) or features? In either case, please provide a description. • What data does each instance consist of? • Is there a label or target associated with each instance? If so, please provide a description. • Is there a label or target associated with each instance? • Is any information missing from individual instances? If so, please provide a description, explaining why this information is missing (e.g., because it was unavailable). This does not include intentionally removed information, but might include, e.g., redacted text. • Is any information missing from individual instances? • Are relationships between individual instances made explicit (e.g., users’ movie ratings, social network links)? If so, please describe how these relationships are made explicit. • Are relationships between individual instances made explicit (e.g., users’ movie ratings, social network links)? • Are there recommended data splits (e.g., training, development/validation, testing)? If so, please provide a description of these splits, explaining the rationale behind them. • Are there recommended data splits (e.g., training, development/validation, testing)? • Are there any errors, sources of noise, or redundancies in the dataset? If so, please provide a description. • Are there any errors, sources of noise, or redundancies in the dataset? If so, please provide a description. • Is the dataset self-contained, or does it link to or otherwise rely on external resources (e.g., websites, tweets, other datasets)? If it links to or relies on external resources, a) are there guarantees that they will exist, and remain constant, over time; b) are there official archival versions of the complete dataset (i.e., including the external resources as they existed at the time the dataset was created); c) are there any restrictions (e.g., licenses, fees) associated with any of the external resources that might apply to a dataset consumer? Please provide descriptions of all external resources and any restrictions associated with them, as well as links or other access points, as appropriate. • Is the dataset self-contained, or does it link to or otherwise rely on external resources (e.g., websites, tweets, other datasets)? • Does the dataset contain data that might be considered confidential (e.g., data that is protected by legal privilege or by doctor– patient confidentiality, data that includes the content of individuals’ non-public communications)? If so, please provide a description. • Does the dataset contain data that might be considered confidential (e.g., data that is protected by legal privilege or by doctor– patient confidentiality, data that includes the content of individuals’ non-public communications)? • Does the dataset contain data that, if viewed directly, might be offensive, insulting, threatening, or might otherwise cause anxiety? If so, please describe why. • Does the dataset contain data that, if viewed directly, might be offensive, insulting, threatening, or might otherwise cause anxiety? If the dataset does not relate to people, you may skip the remaining questions in this section. • Does the dataset identify any subpopulations (e.g., by age, gender)? If so, please describe how these subpopulations are identified and provide a description of their respective distributions within the dataset. • Does the dataset identify any subpopulations (e.g., by age, gender)? • Is it possible to identify individuals (i.e., one or more natural persons), either directly or indirectly (i.e., in combination with other data) from the dataset? If so, please describe how. • Is it possible to identify individuals (i.e., one or more natural persons), either directly or indirectly (i.e., in combination with other data) from the dataset? • Does the dataset contain data that might be considered sensitive in any way (e.g., data that reveals race or ethnic origins, sexual orientations, religious beliefs, political opinions or union memberships, or locations; financial or health data; biometric or genetic data; forms of government identification, such as social security numbers; criminal history)? If so, please provide a description. • Does the dataset contain data that might be considered sensitive in any way (e.g., data that reveals race or ethnic origins, sexual orientations, religious beliefs, political opinions or union memberships, or locations; financial or health data; biometric or genetic data; forms of government identification, such as social security numbers; criminal history)? • Any other comments? • Any other comments? This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Understanding Dataset Instances and Relationships

Authors:
(1) Kedan Li, University of Illinois at Urbana-Champaign;
(2) Min Jin Chong, University of Illinois at Urbana-Champaign;
(3) Jingen Liu, JD AI Research;
(4) David Forsyth, University of Illinois at Urbana-Champaign. Table of Links Abstract and Intro
Related Work
Proposed Method
Experiments
Conclusions and References 2. Related Work Image synthesis: Spatial transformer networks estimate geometric transformations using neural networks [23]. Subsequent work [28,39] shows how to warp one object to another. Warping can be used to produce images of rigid objects [26,30] and non-rigid objects (e.g., clothing) [17,12,45]. In contrast to prior work, we use multiple spatial warpers. Our warps must be combined into a single image, and our U-Net for producing this image follows trends in inpainting (methods that fill in missing portions of an image, see [48,31,50,49]). Han et al. [16,52] show inpainting methods can complete missing clothing items on people. In our work, we use FID∞ to quantitatively evaluate our method. This is based on the Frchet Inception Distance (FID) [18], a common metric in generative image modelling [5,54,29]. Chong et al. [9] recently showed that FID is biased; extrapolation removes the bias, to an unbiased score (FID∞). Generating clothed people: Zhu et al. [57] used a conditional GAN to generate images based on pose skeleton and text descriptions of garment. SwapNet [38] learns to transfer clothes from person A to person B by disentangling clothing and pose features. Hsiao et al. [20] learned a fashion model synthesis network using per-garment encodings to enable convenient minimal edit to specific items. In contrast, we warp products onto real model images. Shape matching underlies our method to match product to model. Tsiao et al. [19] built a shape embedding to enable matching between human body and well-fitting clothing items. Prior work estimated the shape of human body [4,27], clothing items [10,25] and both [35,40], through 2D images. The DensePose [1] descriptor helps modeling the deformation and shading of cloth and, therefore, has been adopted by recent work [36,13,47,51,7,52]. Virtual try-on (VTO) maps a product to a model image. VITON [17] uses a U-Net to generate a coarse synthesis and a mask on the model where the product is presented. A mapping from the product mask to the on-model mask is learned through Thin plate spline (TPS) transformation [3]. The learned mapping is applied on the product image to create a warp. Following their work, Wang et al. [45] improved the architecture using a Geometric Matching Module [39] to estimate the TPS transformations parameters directly from pairs of product image and target person. They train a separate refinement network to combine the warp and the target image. VTNFP [53] extends the work by incorporatiing body segments prediction and later works follow similar procedure [37,24,42,22,2]. However, TPS transformation fails to produce reasonable warps, due to the noisiness of generated masks in our dataset, as shown in Figure 6 right. Instead, we adopt affine transformations which we have found to be more robust to imperfections instead of TPS transformation. A group of following work extended the task to multi-pose. Warping-GAN [11] combined adversarial training with GMM, and generate post and texture separately using a two stage network. MG-VTON [12] further refine the generation method using a three-stage generation network. Other work [21,55,51,7,46] followed similar procedure. Han et al. [15] argued that TPS transformation has low degree of freedom and proposed a flow-based method to create the warp. Much existing virtual try-on work [17,12,21,47,55,53,24,37] is evaluated on datasets that only have tops (t-shirt, shirt, etc.). Having only tops largely reduces the likelihood of shape mismatch as tops have simple and similar shapes. In our work, we extend the problem to include clothing items of all categories(tshirt, shirt, pants, shorts, dress, skirt, robe, jacket, coat, etc.), and propose a method for matching the shape between the source product and the target model. Evaluation shows that using pairs that match in shape significantly increases the generation quality for both our and prior work (table 4.3). In addition, real studio outfits are often covered by an unzipped/unbuttoned outerwear, which is also not presented in prior work [17,12,21,47,55,53,37]. This can cause partition or severe occlusion to the garment, and is not addressed by prior work as shown in Figure 6. We show that our multi-warp generation module ameliorates these difficulties. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Kedan Li, University of Illinois at Urbana-Champaign; (2) Min Jin Chong, University of Illinois at Urbana-Champaign; (3) Jingen Liu, JD AI Research; (4) David Forsyth, University of Illinois at Urbana-Champaign. Authors: Authors: (1) Kedan Li, University of Illinois at Urbana-Champaign; (2) Min Jin Chong, University of Illinois at Urbana-Champaign; (3) Jingen Liu, JD AI Research; (4) David Forsyth, University of Illinois at Urbana-Champaign. Table of Links Abstract and Intro Related Work Proposed Method Experiments Conclusions and References Abstract and Intro Abstract and Intro Related Work Related Work Proposed Method Proposed Method Experiments Experiments Conclusions and References Conclusions and References 2. Related Work Image synthesis : Spatial transformer networks estimate geometric transformations using neural networks [23]. Subsequent work [28,39] shows how to warp one object to another. Warping can be used to produce images of rigid objects [26,30] and non-rigid objects (e.g., clothing) [17,12,45]. In contrast to prior work, we use multiple spatial warpers. Image synthesis Our warps must be combined into a single image, and our U-Net for producing this image follows trends in inpainting (methods that fill in missing portions of an image, see [48,31,50,49]). Han et al. [16,52] show inpainting methods can complete missing clothing items on people. In our work, we use FID∞ to quantitatively evaluate our method. This is based on the Frchet Inception Distance (FID) [18], a common metric in generative image modelling [5,54,29]. Chong et al. [9] recently showed that FID is biased; extrapolation removes the bias, to an unbiased score (FID∞). Generating clothed people: Zhu et al. [57] used a conditional GAN to generate images based on pose skeleton and text descriptions of garment. SwapNet [38] learns to transfer clothes from person A to person B by disentangling clothing and pose features. Hsiao et al. [20] learned a fashion model synthesis network using per-garment encodings to enable convenient minimal edit to specific items. In contrast, we warp products onto real model images. Generating clothed people: Shape matching underlies our method to match product to model. Tsiao et al. [19] built a shape embedding to enable matching between human body and well-fitting clothing items. Prior work estimated the shape of human body [4,27], clothing items [10,25] and both [35,40], through 2D images. The DensePose [1] descriptor helps modeling the deformation and shading of cloth and, therefore, has been adopted by recent work [36,13,47,51,7,52]. Shape matching Virtual try-on (VTO) maps a product to a model image. VITON [17] uses a U-Net to generate a coarse synthesis and a mask on the model where the product is presented. A mapping from the product mask to the on-model mask is learned through Thin plate spline (TPS) transformation [3]. The learned mapping is applied on the product image to create a warp. Following their work, Wang et al. [45] improved the architecture using a Geometric Matching Module [39] to estimate the TPS transformations parameters directly from pairs of product image and target person. They train a separate refinement network to combine the warp and the target image. VTNFP [53] extends the work by incorporatiing body segments prediction and later works follow similar procedure [37,24,42,22,2]. However, TPS transformation fails to produce reasonable warps, due to the noisiness of generated masks in our dataset, as shown in Figure 6 right. Instead, we adopt affine transformations which we have found to be more robust to imperfections instead of TPS transformation. A group of following work extended the task to multi-pose. Warping-GAN [11] combined adversarial training with GMM, and generate post and texture separately using a two stage network. MG-VTON [12] further refine the generation method using a three-stage generation network. Other work [21,55,51,7,46] followed similar procedure. Han et al. [15] argued that TPS transformation has low degree of freedom and proposed a flow-based method to create the warp. Virtual try-on Much existing virtual try-on work [17,12,21,47,55,53,24,37] is evaluated on datasets that only have tops (t-shirt, shirt, etc.). Having only tops largely reduces the likelihood of shape mismatch as tops have simple and similar shapes. In our work, we extend the problem to include clothing items of all categories(tshirt, shirt, pants, shorts, dress, skirt, robe, jacket, coat, etc.), and propose a method for matching the shape between the source product and the target model. Evaluation shows that using pairs that match in shape significantly increases the generation quality for both our and prior work (table 4.3). In addition, real studio outfits are often covered by an unzipped/unbuttoned outerwear, which is also not presented in prior work [17,12,21,47,55,53,37]. This can cause partition or severe occlusion to the garment, and is not addressed by prior work as shown in Figure 6. We show that our multi-warp generation module ameliorates these difficulties. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. available on arxiv

Toward Accurate, Realistic Virtual Try-on Through Shape Matching: Related Work

শেপ ম্যাচিং এর মাধ্যমে সঠিক, বাস্তবসম্মত ভার্চুয়াল ট্রাই-অন: সম্পর্কিত কাজ

Auf dem Weg zu präzisem, realistischem virtuellem Anprobieren durch Formabgleich: Verwandte Arbeiten

Hacia una prueba virtual precisa y realista mediante la combinación de formas: trabajo relacionado

Vers un essai virtuel précis et réaliste grâce à la correspondance de formes : travaux connexes

आकार मिलान के माध्यम से सटीक, यथार्थवादी आभासी ट्राई-ऑन की ओर: संबंधित कार्य

形状マッチングによる正確でリアルな仮想試着に向けて：関連研究

형태 매칭을 통한 정확하고 실감나는 가상 시착을 향하여: 관련 연구

Rumo a um teste virtual preciso e realista por meio da correspondência de formas: trabalho relacionado

На пути к точной и реалистичной виртуальной примерке посредством сопоставления форм: сопутствующие работы

Şekil Eşleştirme Yoluyla Doğru, Gerçekçi Sanal Denemeye Doğru: İlgili Çalışma

Hướng tới việc thử nghiệm ảo thực tế, chính xác thông qua việc khớp hình dạng: Công việc liên quan

通过形状匹配实现准确、逼真的虚拟试穿：相关工作

Authors:
(1) Juan F. Montesinos, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {juanfelipe.montesinos@upf.edu};
(2) Olga Slizovskaia, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {olga.slizovskaia@upf.edu};
(3) Gloria Haro, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {gloria.haro@upf.edu}. Table of Links Abstract and Intro
Related work
Dataset
Experiments
Conclusions and References II. RELATED WORK The University of Rochester Multi-Modal Music Performance Dataset (URMP) [1] is a dataset with 44 Multi-instrument video recordings of classical music pieces. Each instrument present in a piece was recorded separately, both with video and high-quality audio with a stand-alone microphone, in order to have ground-truth individual tracks. Although playing separately, the instruments were coordinated by using a conducting video with a pianist playing in order to set the common timing for the different players. After synchronization, the audio of the individual videos was replaced by the high-quality audio of the microphone and then different recordings were assembled to create the mixture: the individual high-quality audio recordings were added up to create the audio mixture and the visual content was composited in a single video with a common background where all players were arranged at the same level from left to right. For each piece, the dataset provides the musical score in MIDI format, the high-quality individual instrument audio recordings and the videos of the assembled pieces. The instruments present in the dataset, shown in Figure 1, are common instruments in chamber orchestras. In spite of all its good characteristics, it is a small dataset and thus not appropriate for training deep learning architectures. Two other datasets of audio-visual recordings of musical instruments performances have been presented recently: Music [23] and MusicES [31]. Music consists of 536 recordings of solos and 149 videos of duets across 11 categories: accordion, acoustic guitar, cello, clarinet, erhu, flute, saxophone, trumpet, tuba, violin and xylophone. This dataset was gathered by querying YouTube. MusicES [31] is an extension of MUSIC to around the triple of its original size with approximately 1475 recordings but spread in 9 categories instead: accordion, guitar, cello, flute, saxophone, trumpet, tuba, violin and xylophone. There are 7 common categories in MUSIC and Solos: violin, cello, flute, clarinet, saxophone, trumpet and tuba. The common categories between MusicES and Solos are 6 (the former ones except clarinet). Solos and MusicES are complementary. There is only an small intersection of 5% between both, which means both datasets can be combined into a bigger one. We can find in the literature several examples which show the utility of audio-visual datasets. The Sound of Pixels [23] performs audio source separation generating audio spectral components which are further smartly selected by using visual features coming from the video stream to obtain separated sources. This idea was further extended in [20] in order to separate the different sounds present in the mixture in a recursive way. At each stage, the system separates the most salient source from the ones remaining in the mixture. The Sound of Motions [19] uses dense trajectories obtained from optical flow to condition audio source separation, being able even to separate same-instrument mixtures. Visual conditioning is also used in [18] to separate different instruments; during training, a classification loss is used on the separated sounds to enforce object consistency and a co-separation loss forces the estimated individual sounds to produce the original mixtures once reassembled. In [17], the authors developed an energy-based method which minimizes a Non-Negative Matrix Factorization term with an activation matrix which is forced to be aligned to a matrix containing per-source motion information. This motion matrix contains the average magnitude velocities of the clustered motion trajectories in each player bounding box. Recent works show the rising use of skeletons in audiovisual tasks. In Audio to body dynamics [29] authors show it is possible to predict skeletons reproducing the movements of players playing instruments such as piano or violin. Skeletons have proven to be useful for establishing audio-visual correspondences, such as body or finger motion with note onsets or pitch fluctuations, in chamber music performances [21]. A recent work [32] tackles the source separation problem in a similar to Sound of Motions [19] but replacing the dense trajectories by skeleton information. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Juan F. Montesinos, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {juanfelipe.montesinos@upf.edu}; (2) Olga Slizovskaia, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {olga.slizovskaia@upf.edu}; (3) Gloria Haro, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {gloria.haro@upf.edu}. Authors: Authors: (1) Juan F. Montesinos, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {juanfelipe.montesinos@upf.edu}; (2) Olga Slizovskaia, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {olga.slizovskaia@upf.edu}; (3) Gloria Haro, Department of Information and Communications Technologies Universitat Pompeu Fabra, Barcelona, Spain {gloria.haro@upf.edu}. Table of Links Abstract and Intro Related work Dataset Experiments Conclusions and References Abstract and Intro Abstract and Intro Related work Related work Dataset Dataset Experiments Experiments Conclusions and References Conclusions and References II. RELATED WORK The University of Rochester Multi-Modal Music Performance Dataset (URMP) [1] is a dataset with 44 Multi-instrument video recordings of classical music pieces. Each instrument present in a piece was recorded separately, both with video and high-quality audio with a stand-alone microphone, in order to have ground-truth individual tracks. Although playing separately, the instruments were coordinated by using a conducting video with a pianist playing in order to set the common timing for the different players. After synchronization, the audio of the individual videos was replaced by the high-quality audio of the microphone and then different recordings were assembled to create the mixture: the individual high-quality audio recordings were added up to create the audio mixture and the visual content was composited in a single video with a common background where all players were arranged at the same level from left to right. For each piece, the dataset provides the musical score in MIDI format, the high-quality individual instrument audio recordings and the videos of the assembled pieces. The instruments present in the dataset, shown in Figure 1, are common instruments in chamber orchestras. In spite of all its good characteristics, it is a small dataset and thus not appropriate for training deep learning architectures. Two other datasets of audio-visual recordings of musical instruments performances have been presented recently: Music [23] and MusicES [31]. Music consists of 536 recordings of solos and 149 videos of duets across 11 categories: accordion, acoustic guitar, cello, clarinet, erhu, flute, saxophone, trumpet, tuba, violin and xylophone. This dataset was gathered by querying YouTube. MusicES [31] is an extension of MUSIC to around the triple of its original size with approximately 1475 recordings but spread in 9 categories instead: accordion, guitar, cello, flute, saxophone, trumpet, tuba, violin and xylophone. There are 7 common categories in MUSIC and Solos: violin, cello, flute, clarinet, saxophone, trumpet and tuba. The common categories between MusicES and Solos are 6 (the former ones except clarinet). Solos and MusicES are complementary. There is only an small intersection of 5% between both, which means both datasets can be combined into a bigger one. We can find in the literature several examples which show the utility of audio-visual datasets. The Sound of Pixels [23] performs audio source separation generating audio spectral components which are further smartly selected by using visual features coming from the video stream to obtain separated sources. This idea was further extended in [20] in order to separate the different sounds present in the mixture in a recursive way. At each stage, the system separates the most salient source from the ones remaining in the mixture. The Sound of Motions [19] uses dense trajectories obtained from optical flow to condition audio source separation, being able even to separate same-instrument mixtures. Visual conditioning is also used in [18] to separate different instruments; during training, a classification loss is used on the separated sounds to enforce object consistency and a co-separation loss forces the estimated individual sounds to produce the original mixtures once reassembled. In [17], the authors developed an energy-based method which minimizes a Non-Negative Matrix Factorization term with an activation matrix which is forced to be aligned to a matrix containing per-source motion information. This motion matrix contains the average magnitude velocities of the clustered motion trajectories in each player bounding box. Recent works show the rising use of skeletons in audiovisual tasks. In Audio to body dynamics [29] authors show it is possible to predict skeletons reproducing the movements of players playing instruments such as piano or violin. Skeletons have proven to be useful for establishing audio-visual correspondences, such as body or finger motion with note onsets or pitch fluctuations, in chamber music performances [21]. A recent work [32] tackles the source separation problem in a similar to Sound of Motions [19] but replacing the dense trajectories by skeleton information. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. available on arxiv

Solos: A Dataset for Audio-Visual Music Analysis - Related work

একক: অডিও-ভিজ্যুয়াল সঙ্গীত বিশ্লেষণের জন্য একটি ডেটাসেট - সম্পর্কিত কাজ

Solos: Ein Datensatz für audiovisuelle Musikanalyse - Verwandte Arbeiten

Solos: un conjunto de datos para el análisis de música audiovisual - Trabajo relacionado

Solos : un ensemble de données pour l'analyse de la musique audiovisuelle - Travaux connexes

सोलोस: ऑडियो-विज़ुअल संगीत विश्लेषण के लिए एक डेटासेट - संबंधित कार्य

Solos: オーディオビジュアル音楽分析のためのデータセット - 関連研究

Solos: 시청각 음악 분석을 위한 데이터세트 - 관련 작업

Solos: um conjunto de dados para análise musical audiovisual - Trabalhos relacionados

Соло: набор данных для аудиовизуального музыкального анализа — соответствующая работа

Sololar: Görsel-İşitsel Müzik Analizi için Bir Veri Kümesi - İlgili çalışma

Solos: Bộ dữ liệu để phân tích âm nhạc nghe nhìn - Công việc liên quan

Solos：视听音乐分析数据集 - 相关工作

Author:
(1) David S. Hardin, Cedar Rapids, IA USA david.s.hardin@gmail.com. Table of Links Abstract and Introduction
Dancing Links
The Rust Programming Language
RAC: Hardware/Software Co-Assurance at Scale
Rust and RAR
Dancing Links in Rust
Related Work
Conclusion, Acknowledgments, and References 7 Related Work A number of domain-specific languages targeting both hardware and software realization, and providing support for formal verification, have been created. Cryptol [5], for example, has been employed as a “golden spec” for the evaluation of cryptographic implementations, in which automated tools perform equivalence checking between the Cryptol spec for a given algorithm, and the VHDL implementation. Formal verification systems for Rust include Creusot [8], based on WhyML; Prusti [3], based on the Viper verification toolchain; and RustHorn [20], based on constrained Horn clauses. AWS is developing a model-checker for Rust, Kani [2]. Additionally, Carnegie-Mellon University is developing Verus, an SMT-based tool for formally verifying Rust programs [19]. With Verus, programmers express proofs and specifications using Rust syntax, allowing proofs to take advantage of Rust’s linear types and borrow checking. It will be interesting to attempt the sorts of correctness proofs achievable on our system using these verification tools This paper is available on arxiv under CC 4.0 license. Author: (1) David S. Hardin, Cedar Rapids, IA USA david.s.hardin@gmail.com. Author: Author: (1) David S. Hardin, Cedar Rapids, IA USA david.s.hardin@gmail.com. Table of Links Abstract and Introduction Dancing Links The Rust Programming Language RAC: Hardware/Software Co-Assurance at Scale Rust and RAR Dancing Links in Rust Related Work Conclusion, Acknowledgments, and References Abstract and Introduction Abstract and Introduction Dancing Links Dancing Links The Rust Programming Language The Rust Programming Language RAC: Hardware/Software Co-Assurance at Scale RAC: Hardware/Software Co-Assurance at Scale Rust and RAR Rust and RAR Dancing Links in Rust Dancing Links in Rust Related Work Related Work Conclusion, Acknowledgments, and References Conclusion, Acknowledgments, and References 7 Related Work A number of domain-specific languages targeting both hardware and software realization, and providing support for formal verification, have been created. Cryptol [5], for example, has been employed as a “golden spec” for the evaluation of cryptographic implementations, in which automated tools perform equivalence checking between the Cryptol spec for a given algorithm, and the VHDL implementation. Formal verification systems for Rust include Creusot [8], based on WhyML; Prusti [3], based on the Viper verification toolchain; and RustHorn [20], based on constrained Horn clauses. AWS is developing a model-checker for Rust, Kani [2]. Additionally, Carnegie-Mellon University is developing Verus, an SMT-based tool for formally verifying Rust programs [19]. With Verus, programmers express proofs and specifications using Rust syntax, allowing proofs to take advantage of Rust’s linear types and borrow checking. It will be interesting to attempt the sorts of correctness proofs achievable on our system using these verification tools This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv

Verification of a Rust Implementation of Knuth’s Dancing
Links Using ACL2: Related Work

Stories, News & Mentions about RELLA

Rella’s Stories on HackerNoon