klamp

Startups of The Year 2023 Winners: North America 

This paper is under CC BY 4.0 DEED license. available on arxiv Authors: (1) Ehsan Toreini, University of Surrey, UK; (2) Maryam Mehrnezhad, Royal Holloway University of London; (3) Aad Van Moorsel, Birmingham University. Table of Links Abstract & Introduction Background and Related Work FaaS Architecture Implementation and Performance Analysis Conclusion Acknowledgment and References Acknowledgment The authors in this project have been funded by UK EPSRC grant “FinTrust: Trust Engineering for the Financial Industry” under grant number EP/R033595/1, and UK EPSRC grant “AGENCY: Assuring Citizen Agency in a World with Complex Online Harms” under grant EP/W032481/1 and PETRAS National Centre of Excellence for IoT Systems Cybersecurity, which has been funded by the UK EPSRC under grant number EP/S035362/1. References [1] Philip Adler, Casey Falk, Sorelle A Friedler, Tionney Nix, Gabriel Rybeck, Carlos Scheidegger, Brandon Smith, and Suresh Venkatasubramanian. Auditing black-box models for indirect influence. Knowledge and Information Systems, 54(1):95–122, 2018. [2] Elaine Angelino, Nicholas Larus-Stone, Daniel Alabi, Margo Seltzer, and Cynthia Rudin. Learning certifiably optimal rule lists for categorical data. arXiv preprint arXiv:1704.01701, 2017. [3] Muhammad Ajmal Azad, Samiran Bag, Simon Parkinson, and Feng Hao. Trustvote: Privacy-preserving node ranking in vehicular networks. IEEE Internet of Things Journal, 6(4):5878–5891, 2018. [4] Olivier Baudron, Pierre-Alain Fouque, David Pointcheval, Jacques Stern, and Guillaume Poupard. Practical multi-candidate election system. In Proceedings of the twentieth annual ACM symposium on Principles of distributed computing, pages 274–283, 2001. [5] Alexandra Chouldechova. Fair prediction with disparate impact: A study of bias in recidivism prediction instruments. Big data, 5(2):153–163, 2017. [6] Sam Corbett-Davies, Emma Pierson, Avi Feller, Sharad Goel, and Aziz Huq. Algorithmic decision making and the cost of fairness. In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 797–806. ACM, 2017. [7] Ronald Cramer, Ivan Damg˚ard, and Berry Schoenmakers. Proofs of partial knowledge and simplified design of witness hiding protocols. In Annual International Cryptology Conference, pages 174–187. Springer, 1994. [8] Craig E. Carroll and Rowena Olegario. Pathways to corporate accountability: Corporate reputation and its alternatives. Journal of Business Ethics, 163(2):173–181, 2020. [9] Michael Feldman, Sorelle A Friedler, John Moeller, Carlos Scheidegger, and Suresh Venkatasubramanian. Certifying and removing disparate impact. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 259–268. ACM, 2015. [10] Amos Fiat and Adi Shamir. How to prove yourself: Practical solutions to identification and signature problems. In Conference on the theory and application of cryptographic techniques, pages 186–194. Springer, 1986. [11] Sorelle A Friedler, Carlos Scheidegger, and Suresh Venkatasubramanian. On the (im) possibility of fairness. arXiv preprint arXiv:1609.07236, 2016. [12] Riccardo Guidotti, Anna Monreale, Salvatore Ruggieri, Franco Turini, Fosca Giannotti, and Dino Pedreschi. A survey of methods for explaining black box models. ACM computing surveys (CSUR), 51(5):1–42, 2018. [13] Feng Hao, Matthew N Kreeger, Brian Randell, Dylan Clarke, Siamak F Shahandashti, and Peter Hyun-Jeen Lee. Every vote counts: Ensuring integrity in large-scale electronic voting. In 2014 Electronic Voting Technology Workshop/Workshop on Trustworthy Elections (EVT/WOTE 14), 2014. [14] Feng Hao, Peter YA Ryan, and Piotr Zieli´nski. Anonymous voting by two-round public discussion. IET Information Security, 4(2):62–67, 2010. [15] Moritz Hardt, Eric Price, Nati Srebro, and others. Equality of opportunity in supervised learning. In Advances in neural information processing systems, pages 3315–3323, 2016. [16] Hui Hu, Yijun Liu, Zhen Wang, and Chao Lan. A distributed fair machine learning framework with private demographic data protection. In 2019 IEEE International Conference on Data Mining (ICDM), pages 1102–1107. IEEE, 2019. [17] Matthew Jagielski, Michael Kearns, Jieming Mao, Alina Oprea, Aaron Roth, Saeed Sharifi-Malvajerdi, and Jonathan Ullman. Differentially private fair learning. In International Conference on Machine Learning, pages 3000–3008. PMLR, 2019. [18] Jonathan Katz and Yehuda Lindell. Introduction to modern cryptography. CRC press, 2014. [19] Niki Kilbertus, Adria Gascon, Matt Kusner, Michael Veale, Krishna P Gummadi, and Adrian Weller. Blind justice: Fairness with encrypted sensitive attributes. In 35th International Conference on Machine Learning, pages 2630–2639. PMLR, 2018. [20] Jeff Larson, Surya Mattu, Lauren Kirchner, and Julia Angwin. How we analyzed the COMPAS recidivism algorithm. ProPublica (5 2016), 9(1), 2016. [21] Jiacheng Liu, Fei Yu, and Lixin Song. A systematic investigation on the research publications that have used the medical expenditure panel survey (MEPS) data through a bibliometrics approach. Library Hi Tech, 2020. [22] Scott M Lundberg and Su-In Lee. A unified approach to interpreting model predictions. Advances in neural information processing systems, 30, 2017. [23] Arwa Mahdawi. It’s not just A-levels – algorithms have a nightmarish new power over our lives. The Guardian, 2020. [24] Arvind Narayanan. Translation tutorial: 21 fairness definitions and their politics. In Proc. Conf. Fairness Accountability Transp., New York, USA, 2018. [25] Cecilia Panigutti, Alan Perotti, Andr´e Panisson, Paolo Bajardi, and Dino Pedreschi. Fairlens: Auditing black-box clinical decision support systems. Information Processing & Management, 58(5):102657, 2021. [26] Cecilia Panigutti, Alan Perotti, and Dino Pedreschi. Doctor xai: an ontology-based approach to black-box sequential data classification explanations. In Proceedings of the 2020 conference on fairness, accountability, and transparency, pages 629–639, 2020. [27] Saerom Park, Seongmin Kim, and Yeon-sup Lim. Fairness audit of machine learning models with confidential computing. In Proceedings of the ACM Web Conference 2022, pages 3488–3499, 2022. [28] Reuters. Amazon ditched AI recruiting tool that favored men for technical jobs. The Guardian, 2018. [29] Shahar Segal, Yossi Adi, Benny Pinkas, Carsten Baum, Chaya Ganesh, and Joseph Keshet. Fairness in the eyes of the data: Certifying machinelearning models. In Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society, pages 926–935, 2021. [30] Keng Siau and Weiyu Wang. Building trust in artificial intelligence, machine learning, and robotics. Cutter business technology journal, 31(2):47– 53, 2018. [31] Douglas Robert Stinson and Maura Paterson. Cryptography: theory and practice. CRC press, 2018. [32] Ehsan Toreini, Mhairi Aitken, Kovila Coopamootoo, Karen Elliott, Carlos Gonzalez Zelaya, and Aad van Moorsel. The relationship between trust in AI and trustworthy machine learning technologies. In Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency, pages 272–283, 2020. [33] Michael Veale and Reuben Binns. Fairer machine learning in the real world: Mitigating discrimination without collecting sensitive data. Big Data & Society, 4(2):2053951717743530, 2017. [34] Tong Wang, Cynthia Rudin, Finale Doshi-Velez, Yimin Liu, Erica Klampfl, and Perry MacNeille. A bayesian framework for learning rule sets for interpretable classification. The Journal of Machine Learning Research, 18(1):2357–2393, 2017. [35] Qiang Yang, Yang Liu, Tianjian Chen, and Yongxin Tong. Federated machine learning: Concept and applications. ACM Transactions on Intelligent Systems and Technology (TIST), 10(2):1–19, 2019.

Privacy-preserving Computation of Fairness for ML Systems: Acknowledgement & References

KUOPIO, Finland, September 12th, 2024/Chainwire/--Kampela, an innovative hardware startup, has successfully secured funding through a Decentralized Autonomous Organization (DAO) on the Polkadot network, marking a significant milestone as the first hardware project to be fully backed by a DAO in the Polkadot ecosystem. This achievement demonstrates the power of decentralized governance in supporting tangible technological innovations and opens new possibilities for hardware development in the blockchain space. Kampela's device leverages NFC technology and operates without the need for wired charging or a battery, showcasing a novel approach to secure, user-centric hardware wallets. The company's innovative strategy consolidates both the design and production processes within a single integrated platform, enabling the delivery of highly differentiated solutions to the market. Key Highlights: Funding secured through Polkadot Referenda #62, #370, and #886, totalling approximately 253,000 DOTs (around $1 million USD at current DOT price)
Transparent, community-driven decision-making process
Establishes a new model for supporting hardware innovation in the blockchain industry
Battery-free operation powered exclusively by NFC during active use
Secure NFC-only connectivity, eliminating vulnerable USB interfaces Kirill, ex-CISO of Parity and Kampela's Co-Founder, stated, "Not only is Kampela one of the very few fully open hardware wallets on the markets (you can get everything, up to the precise machining instructions for the casing, from our GitHub), but the project is also a brilliant example of the power of DAOs in the new, blockchain-enabled economy: the project was funded via the Polkadot Treasury (one of the biggest DAOs in the world right now), and it is thanks to this funding that our designs are so uncompromisingly open, free, and democratic." The successful funding of Kampela through a DAO on the Polkadot network showcases how community-driven funding can fuel advancements in secure, user-centric devices. This achievement is expected to attract attention from blockchain enthusiasts, hardware innovators, investors interested in decentralized funding models, and Polkadot community members and stakeholders. Kampela's hardware design features several enhanced security measures, including battery-free operation and secure NFC-only connectivity. The device remains completely powered down when not in use, deterring physical access or tampering, and its NFC-only communication provides an additional security layer compared to traditional USB interfaces. With six hardware revisions conducted and 54 components in its design, Kampela represents a significant leap forward in blockchain hardware innovation. As the first fully DAO-funded hardware project in the Polkadot ecosystem, Kampela sets a new precedent for how hardware startups can leverage decentralized funding and governance to bring cutting-edge products to market.
For more information about Kampela and its innovative hardware wallet, readers can visit https://www.kampe.la/ or https://x.com/kampela_signer. About Kampela Kampela, a Finnish startup, is revolutionizing secure transaction management with its innovative NFC-powered hardware device. Operating without Bluetooth, USB, or batteries, it features an e-ink display and open-source design. Tailored for blockchain applications and backed by the Web3 Foundation and Polkadot Treasury, Kampela offers a sustainable, secure solution for digital interactions in decentralized systems. Media Contact: Viktoria CMO [Telegram: @vikrpc ] Contact CMO Viktoria Kampela vv.repich@gmail.com This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here. KUOPIO, Finland, September 12th, 2024/Chainwire/--Kampela, an innovative hardware startup, has successfully secured funding through a Decentralized Autonomous Organization (DAO) on the Polkadot network, marking a significant milestone as the first hardware project to be fully backed by a DAO in the Polkadot ecosystem. This achievement demonstrates the power of decentralized governance in supporting tangible technological innovations and opens new possibilities for hardware development in the blockchain space. Kampela's device leverages NFC technology and operates without the need for wired charging or a battery, showcasing a novel approach to secure, user-centric hardware wallets. The company's innovative strategy consolidates both the design and production processes within a single integrated platform, enabling the delivery of highly differentiated solutions to the market. Key Highlights: Funding secured through Polkadot Referenda #62, #370, and #886, totalling approximately 253,000 DOTs (around $1 million USD at current DOT price) Transparent, community-driven decision-making process Establishes a new model for supporting hardware innovation in the blockchain industry Battery-free operation powered exclusively by NFC during active use Secure NFC-only connectivity, eliminating vulnerable USB interfaces Funding secured through Polkadot Referenda #62, #370, and #886, totalling approximately 253,000 DOTs (around $1 million USD at current DOT price) Transparent, community-driven decision-making process Establishes a new model for supporting hardware innovation in the blockchain industry Battery-free operation powered exclusively by NFC during active use Secure NFC-only connectivity, eliminating vulnerable USB interfaces Kirill, ex-CISO of Parity and Kampela's Co-Founder, stated, "Not only is Kampela one of the very few fully open hardware wallets on the markets (you can get everything, up to the precise machining instructions for the casing, from our GitHub), but the project is also a brilliant example of the power of DAOs in the new, blockchain-enabled economy: the project was funded via the Polkadot Treasury (one of the biggest DAOs in the world right now), and it is thanks to this funding that our designs are so uncompromisingly open, free, and democratic." Kirill, ex-CISO of Parity and Kampela's Co-Founder, stated, "Not only is Kampela one of the very few fully open hardware wallets on the markets (you can get everything, up to the precise machining instructions for the casing, from our GitHub), but the project is also a brilliant example of the power of DAOs in the new, blockchain-enabled economy: the project was funded via the Polkadot Treasury (one of the biggest DAOs in the world right now), and it is thanks to this funding that our designs are so uncompromisingly open, free, and democratic." The successful funding of Kampela through a DAO on the Polkadot network showcases how community-driven funding can fuel advancements in secure, user-centric devices. This achievement is expected to attract attention from blockchain enthusiasts, hardware innovators, investors interested in decentralized funding models, and Polkadot community members and stakeholders. Kampela's hardware design features several enhanced security measures, including battery-free operation and secure NFC-only connectivity. The device remains completely powered down when not in use, deterring physical access or tampering, and its NFC-only communication provides an additional security layer compared to traditional USB interfaces. With six hardware revisions conducted and 54 components in its design, Kampela represents a significant leap forward in blockchain hardware innovation. As the first fully DAO-funded hardware project in the Polkadot ecosystem, Kampela sets a new precedent for how hardware startups can leverage decentralized funding and governance to bring cutting-edge products to market.
For more information about Kampela and its innovative hardware wallet, readers can visit https://www.kampe.la/ or https://x.com/kampela_signer . https://www.kampe.la/ https://x.com/kampela_signer About Kampela Kampela, a Finnish startup, is revolutionizing secure transaction management with its innovative NFC-powered hardware device. Operating without Bluetooth, USB, or batteries, it features an e-ink display and open-source design. Tailored for blockchain applications and backed by the Web3 Foundation and Polkadot Treasury, Kampela offers a sustainable, secure solution for digital interactions in decentralized systems. Media Contact: Viktoria CMO [Telegram: @vikrpc ] Contact CMO Viktoria Kampela vv.repich@gmail.com vv.repich@gmail.com This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here. This story was distributed as a release by Chainwire under HackerNoon’s Business Blogging Program. Learn more about the program here . here here

DAO-Funded Hardware Wallet Kampela Secures Polkadot Network Investment

DAO-befondsde hardeware-beursie Kampela verseker Polkadot-netwerkbelegging

محفظة الأجهزة Kampela الممولة من DAO تؤمن استثمارًا لشبكة Polkadot

DAO-n qullqichata Hardware Wallet Kampela ukax Polkadot Red ukan Inversión ukarux askinchawayiwa

La billetera de hardware financiada por DAO, Kampela, obtiene inversión en la red Polkadot

DAOk finantzatutako Hardware Wallet Kampelak Polkadot sareko inbertsioa ziurtatzen du

DAO:n rahoittama laitteistolompakko Kampela turvaa Polkadot-verkkosijoituksen

A carteira de hardware financiada por DAO Kampela asegura o investimento na rede de Polkadot

DAO 出資のハードウェアウォレット Kampela が Polkadot ネットワークへの投資を獲得

DAO 자금 지원 하드웨어 지갑 Kampela, Polkadot 네트워크 투자 확보

DAO finansuojama aparatinės įrangos piniginė „Kampela“ apsaugo „Polkadot“ tinklo investicijas

د DAO لخوا تمویل شوي هارډویر والټ کمپیلا د پولکاډوټ شبکې پانګونه خوندي کوي

Kampela กระเป๋าสตางค์ฮาร์ดแวร์ที่ได้รับทุนจาก DAO ได้รับการลงทุนจากเครือข่าย Polkadot

HackerNoon editorial team has launched this interview series with women in tech to celebrate their achievements and share their struggles. We need more women in technology, and by sharing stories, we can encourage many girls to follow their dreams. Share your story today! Tell us about yourself! I am a final year doctoral student at the Laboratory for Computational Social Systems based out of IIIT-Delhi, India. By employing computational tools, mainly applied Natural Language Processing (NLP) and Network Science, I investigate the contextual signals that can help improve the detection of hate speech on the web. I have been fortunate to receive financial aid via the Google Phd Fellowship and the Prime Minister Doctoral Fellowship (in partnership with Wipro AI) to support my studies. Now, towards the end of my Phd, I am expanding my horizon to investigate other forms of harmful content on the web. Lately, I have been exploring the themes of fact-checking, fake news, propaganda analysis, and news narrative understanding. The last of which is the topic of my upcoming postdoc work  at CopeNLU  at the University of Copenhagen. As part of my academic and professional journey, I have been fortunate to work full time at Red Hat, furnish my undergrads in CSE from Jamia Millia Islamia (JMI), be a visiting research student at TUM and volunteer with organizations such as AnitaB and Journal of Open Source Softwares (JOSS). Why did you choose this field in the first place? The idea of pursuing computer science or being a phd student was not something I had planned for. Early exposure to science and technology both at home and school, as well as my curiosity to understand how things work, led me to develop a love of science and engineering. However, once I started my undergraduate studies, I struggled to find exactly what I planned to apply my CSE knowledge towards. I have failed at more undertakings than I have succeeded in. I chose to continue my work and studies in computer science because of its interdisciplinary application, especially once I discovered that I enjoyed NLP as a subfield while doing my undergraduate thesis on topic modelling. My interest in applied NLP was further solidified when I got an opportunity to work in the area of computational social systems (CSS) as a part of my PhD. What was the biggest setback/failure that you faced, and how did you manage it? PhD in itself is full of setbacks and failures, with success sprinkled here and there. One incident I especially remember, as it lasted almost two years, was my work on developing a dataset and model to detect hate speech in codemixed Hinglish. It took us one year to curate and annotate the dataset from Twitter. Given the Indian context and language knowledge needed to detect hateful connotations correctly, we could not employ existing crowdsourcing platforms like  Amazon Mechanical Turk. We spend endless hours performing annotations ourselves to produce an annotation guideline and then seek Indian annotators to help perform large-scale annotations. This process was iterative, slow, and mentally taxing for my colleagues and me. However, our woes did not end there, as the standard practice of finetuning a language model (LM) for hate speech detection did not produce effective results. It took us multiple trials and errors to realise that instead of focusing on building a more "complex and fancy system", we needed to improve the input signals, and it took us another round of unsupervised auxiliary data curation and modelling to augment the standard finetuning pipeline to yield improvements when detecting hatefulness in Indic context. The two years that it took from start to finish of this work were full of uncertainty; coupled with the Covid lockdown, it made me feel as if my phd journey would come to an abrupt end. What got us out of the slump was revisiting the research around contextual information, which was not a common practice in NLP modelling for hate speech but was closer to how content moderators make decisions in the real world. What's your biggest achievement that you're really proud of? One project that I am really proud of is a work we recently published on Arxiv (it is still under review for publication). The project explores the use of GenAI tools like ChatGPT in understanding journalistic narratives of the 5W1H (What, Why, When, Where, Who and How) and uncovering neutrality or lack thereof in independent fact-checking. The research not only reiterates the need to look beyond the USA-centric left-right political bias but also contributes to the initial interdisciplinary of combining human-AI efforts to study large-scale corpus and quantify subtle reporting/narrative biases that go unnoticed. The project idea came to me when I was randomly scrolling through an independent news portal, which is known to be "right-aligned", and saw that they had a very active fact-checking project. It led to framing the question of how explicit or implicit political leaning of the independent fact-checking organizations might lead to deviation from neutrality in fact-checking. I was very happy when what I hypothesised from initial manual observations was successfully quantified by our experiments. The bigger question this leaves for the research community is how we quantify the impact of different versions of "truth" and the subtle biases that cause deviation from a neutral point of view. What tech are you most excited/passionate about right now and why? While working with a lot of Hindi and Hinglish datasets from Twitter for analysing hate speech, I kept coming across gaps in the performance of NLP models when working with non-English text and non-Western cultural references. In this regard, I am excited about exploring both modelling paradigms and evaluation benchmarks to improve the adaptability of current NLP systems to non-Western connotations. Of course, by developing more culturally tuned systems, we will be able to understand toxic and stereotyped references better. I think a safer deployment of NLP-based models and interfaces is only possible if, at the granular level, we actively work on improving their adaptability to specific cultures and linguistic patterns. Another area within CSS that I am keen on exploring is narrative building. I recently read "The Language of the Third Reich" by Victor Klemperer and was amazed by how many of the techniques and patterns of narrative building applied by today's media can be traced back to World War II and prior eras. Given that the human psyche remains more or less the same, no matter the language or medium of storytelling, I am keen on applying NLP to help establish long-range narrative understanding. What tech are you most worried about right now and why? It is not technology per se, but I am worried about AI being presented as either an impeccable panacea that will fix all humanity's problems or an existential threat that will wipe off the human race. Inflating the abilities of AI  beyond what is possible by current paradigms is leading to the investment of resources in the wrong direction. It takes the conversation away from current threats to deploying these systems without proper safety checks in place and ignores the biases it propagates. A  person of colour being wrongly convicted of a crime due to biased AI modelling is much more probable and very much a reality than AGI wiping off humanity. In your opinion, why do we see a huge gender gap in the tech industry, and how can we reduce it? Working in tech, whether in industry or academia, is a very consuming process. On the one hand,  technical skills and the accompanying soft skills are something anyone can acquire via varying training channels (formal or informal). On the other hand, the on-paper equality of the tech space is marred by differences in social expectations from one gender specifically. Unless our social systems become more accommodating of working women and provide adequate moral, financial and infrastructural support, women in tech or the workforce in general will suffer. Do you have any advice for aspiring girls who want to join the field? For girls who aspire to work in tech, I would say that they should not feel intimidated; the tech world is full of possibilities and opportunities. There is room for different projects and perspectives to thrive, so keep participating in as many projects as possible and develop your sense of what works and what does not. What are your hobbies and interests outside of tech? Outside of tech, I really enjoy visiting museums and historical sites. Whenever I find time on weekends, I participate in guided tours in my city. I also dabble in writing poetry sometimes; I like to refer to myself as a seasonal poet :P However, my favourite hobby has always been reading. From books to blogs to research papers, the act of reading is very relaxing to me. HackerNoon editorial team has launched this interview series with women in tech to celebrate their achievements and share their struggles. We need more women in technology, and by sharing stories, we can encourage many girls to follow their dreams. Share your story today ! this interview series with women in tech Share your story today Tell us about yourself! Tell us about yourself! I am a final year doctoral student at the Laboratory for Computational Social Systems based out of IIIT-Delhi, India. By employing computational tools, mainly applied Natural Language Processing (NLP) and Network Science, I investigate the contextual signals that can help improve the detection of hate speech on the web . I have been fortunate to receive financial aid via the Google Phd Fellowship and the Prime Minister Doctoral Fellowship (in partnership with Wipro AI) to support my studies. Now, towards the end of my Phd, I am expanding my horizon to investigate other forms of harmful content on the web. Lately, I have been exploring the themes of fact-checking, fake news, propaganda analysis, and news narrative understanding. The last of which is the topic of my upcoming postdoc work  at CopeNLU  at the University of Copenhagen. hate speech on the web As part of my academic and professional journey, I have been fortunate to work full time at Red Hat, furnish my undergrads in CSE from Jamia Millia Islamia (JMI), be a visiting research student at TUM and volunteer with organizations such as AnitaB and Journal of Open Source Softwares (JOSS). Why did you choose this field in the first place? Why did you choose this field in the first place? The idea of pursuing computer science or being a phd student was not something I had planned for. Early exposure to science and technology both at home and school, as well as my curiosity to understand how things work, led me to develop a love of science and engineering. However, once I started my undergraduate studies, I struggled to find exactly what I planned to apply my CSE knowledge towards. I have failed at more undertakings than I have succeeded in. I chose to continue my work and studies in computer science because of its interdisciplinary application, especially once I discovered that I enjoyed NLP as a subfield while doing my undergraduate thesis on topic modelling. My interest in applied NLP was further solidified when I got an opportunity to work in the area of computational social systems (CSS) as a part of my PhD. What was the biggest setback/failure that you faced, and how did you manage it? What was the biggest setback/failure that you faced, and how did you manage it? PhD in itself is full of setbacks and failures, with success sprinkled here and there. One incident I especially remember, as it lasted almost two years, was my work on developing a dataset and model to detect hate speech in codemixed Hinglish. It took us one year to curate and annotate the dataset from Twitter. Given the Indian context and language knowledge needed to detect hateful connotations correctly, we could not employ existing crowdsourcing platforms like  Amazon Mechanical Turk. We spend endless hours performing annotations ourselves to produce an annotation guideline and then seek Indian annotators to help perform large-scale annotations. This process was iterative, slow, and mentally taxing for my colleagues and me. However, our woes did not end there, as the standard practice of finetuning a language model (LM) for hate speech detection did not produce effective results. It took us multiple trials and errors to realise that instead of focusing on building a more "complex and fancy system", we needed to improve the input signals, and it took us another round of unsupervised auxiliary data curation and modelling to augment the standard finetuning pipeline to yield improvements when detecting hatefulness in Indic context. The two years that it took from start to finish of this work were full of uncertainty; coupled with the Covid lockdown, it made me feel as if my phd journey would come to an abrupt end. What got us out of the slump was revisiting the research around contextual information, which was not a common practice in NLP modelling for hate speech but was closer to how content moderators make decisions in the real world. What's your biggest achievement that you're really proud of? What's your biggest achievement that you're really proud of? One project that I am really proud of is a work we recently published on Arxiv (it is still under review for publication). The project explores the use of GenAI tools like ChatGPT in understanding journalistic narratives of the 5W1H (What, Why, When, Where, Who and How) and uncovering neutrality or lack thereof in independent fact-checking. The research not only reiterates the need to look beyond the USA-centric left-right political bias but also contributes to the initial interdisciplinary of combining human-AI efforts to study large-scale corpus and quantify subtle reporting/narrative biases that go unnoticed. The project idea came to me when I was randomly scrolling through an independent news portal, which is known to be "right-aligned", and saw that they had a very active fact-checking project. It led to framing the question of how explicit or implicit political leaning of the independent fact-checking organizations might lead to deviation from neutrality in fact-checking. I was very happy when what I hypothesised from initial manual observations was successfully quantified by our experiments. The bigger question this leaves for the research community is how we quantify the impact of different versions of "truth" and the subtle biases that cause deviation from a neutral point of view. What tech are you most excited/passionate about right now and why? While working with a lot of Hindi and Hinglish datasets from Twitter for analysing hate speech, I kept coming across gaps in the performance of NLP models when working with non-English text and non-Western cultural references. In this regard, I am excited about exploring both modelling paradigms and evaluation benchmarks to improve the adaptability of current NLP systems to non-Western connotations. Of course, by developing more culturally tuned systems, we will be able to understand toxic and stereotyped references better. I think a safer deployment of NLP-based models and interfaces is only possible if, at the granular level, we actively work on improving their adaptability to specific cultures and linguistic patterns. Another area within CSS that I am keen on exploring is narrative building. I recently read "The Language of the Third Reich" by Victor Klemperer and was amazed by how many of the techniques and patterns of narrative building applied by today's media can be traced back to World War II and prior eras. Given that the human psyche remains more or less the same, no matter the language or medium of storytelling, I am keen on applying NLP to help establish long-range narrative understanding. What tech are you most worried about right now and why? What tech are you most worried about right now and why? It is not technology per se, but I am worried about AI being presented as either an impeccable panacea that will fix all humanity's problems or an existential threat that will wipe off the human race. Inflating the abilities of AI  beyond what is possible by current paradigms is leading to the investment of resources in the wrong direction. It takes the conversation away from current threats to deploying these systems without proper safety checks in place and ignores the biases it propagates. A  person of colour being wrongly convicted of a crime due to biased AI modelling is much more probable and very much a reality than AGI wiping off humanity. In your opinion, why do we see a huge gender gap in the tech industry, and how can we reduce it? In your opinion, why do we see a huge gender gap in the tech industry, and how can we reduce it? Working in tech, whether in industry or academia, is a very consuming process. On the one hand,  technical skills and the accompanying soft skills are something anyone can acquire via varying training channels (formal or informal). On the other hand, the on-paper equality of the tech space is marred by differences in social expectations from one gender specifically. Unless our social systems become more accommodating of working women and provide adequate moral, financial and infrastructural support, women in tech or the workforce in general will suffer. Do you have any advice for aspiring girls who want to join the field? For girls who aspire to work in tech, I would say that they should not feel intimidated; the tech world is full of possibilities and opportunities. There is room for different projects and perspectives to thrive, so keep participating in as many projects as possible and develop your sense of what works and what does not. What are your hobbies and interests outside of tech? What are your hobbies and interests outside of tech? Outside of tech, I really enjoy visiting museums and historical sites. Whenever I find time on weekends, I participate in guided tours in my city. I also dabble in writing poetry sometimes; I like to refer to myself as a seasonal poet :P However, my favourite hobby has always been reading. From books to blogs to research papers, the act of reading is very relaxing to me.

Meet Sarah Masud, Google Phd Fellow and Final Year Doctoral Student at IIIT-Delhi

AI image manipulation tools have come a long way in the last few years. It’s quite easy nowadays to create beautiful, stylistic portraits of people and animals. But AI models are highly unpredictable. So most tools rely on the user (or some human) to weed out bad generations and find the best one. This is the classic “human in the loop” problem that often plagues AI tools. Turns out, with some clever tricks and careful tuning, you can build a pipeline that reliably works for the vast majority of pets. It is extremely resilient to variations in pose, lighting etc. In this post, I’ll dive deeper into how it works and all the neat little tricks that enable this. Here’s some examples of portraits you can generate with this pipeline. Let’s get started! The Key Ingredients IPAdapter The crux of the technique is the IPAdapter. It’s essentially a way to prompt a model using an image instead of text (it literally stands for Image Prompt Adapter). So instead of taking a text embeddings, it uses an image to get the embeddings. This is super powerful because it can accurately capture the style and structure in an image directly, instead of someone having to translate what they want from an image into text. In our ComfyUI IPAdapter node, we have two inputs, one for style and one for composition. We use a watercolor painting image for style, and feed the original image for composition (since we want to keep the same composition, but change the style). ControlNets Now that we have a way to keep the style consistent, we can turn our attention to faithfully representing the pet. IPAdapters bias heavily towards image quality, and image likeness suffers. So we need to do something to be able to keep the output looking like the same object as the input. The answer to that is ControlNet. ControlNets are another cool technique to provide additional constraints to the image generation process. Using a ControlNet, you can specify constraints in the form of edges, depth, human pose etc. A great feature of ControlNets is that they can be stacked. So you can have an edge controlnet that forces the output to have similar edges as the input, and also a depth controlnet that forces the output to have a similar depth profile. And that’s exactly what I do here. It turns out that controlnets are not just stackable with other controlnets, but they can also work in tandem with IPAdapter mentioned above. And so those are the tools we’ll use for this - IPAdapter with a source image to get the style, ControlNet with canny edge detector to constrain based on edges, and controlnet with depth to constrain based on depth profile. That’s really all you need in terms of techniques, but what I’ve learnt from experimenting with machine learning for production is that a lot of the value of these things comes from taking the time to tune all your parameters perfectly. So I want to talk a little bit about that. Finetuning Have you ever found a model with amazing example outputs, tried it on your own images and realized they look terrible? Often, the only reason for that is that the model hasn’t been finetuned for your images. It can feel like a complete blocker sometimes, because where do we even start with finetuning a pre-trained model?! Here’s what I’ve learnt about the topic. This extends beyond this particular pipeline too, so it’s good knowledge to have generally. ControlNets ControlNets are very powerful, so you have to be careful to control their effect on the output. Luckily, the custom nodes in ComfyUI let us reduce the effect of controlnets, and also stop their effects at any point. So we set the edge detector to 75% strength and make it stop influencing the generation at 75% and the depth detector stops at 30%. The reason we stop them at the end rather than just reducing their strength is that this allows the network to “clean up” any artifacts caused by them in the last few steps without being constrained externally. It just makes the image prettier. So it’s using just it’s training data to make things look as nice as possible, ignoring the edges and depth. The other big thing to tune is the KSampler. There’s a lot of little things going on here, but I’ll just briefly touch on some of them: KSampler - Steps First we have the steps. This is literally the number of times the model will repeatedly run. The more it runs, the more stylized your output gets, and the further away it will get from the original image. The effects of this are often not that obvious, so it’s worth playing around with it. KSampler - CFG Then there’s CFG. I honestly don’t fully understand this, but from its name - Classifier Free Guidance - I’m assuming it controls how much the model is allowed to modify the image unconstrained by the prompts in order to make it look better. This also affects the output image significantly, so is worth playing around with. Denoise Another neat little trick I use here is to start the image generation process with the input image  instead of a blank image, and keep the denoising low. This ensures that the output will look similar in terms of colors and textures. Text Prompt One thing you’ll notice I never mentioned is the text prompt until now! Surprising, since that’s usually the only conditioning that you normally provide to diffusion models. But in this case, we have so many other ways of conditioning that the text prompts usually just get in the way. So in this case, the prompt is literally just “a dog”. I do use the text prompt a little more in some of the more stylized portraits, like the chef dog or the one in the bathroom. Adding accessories In essence, this is more or less just an “AI Filter” that converts images into watercolor portraits. But it’s amazing how flexible this can be. For example, to make the portrait of the dog taking a shower, I literally just put images together in an image editing tool and used that as the input! The model takes care of unifying everything and cleaning up the image. Conclusion Now remove the background, add some text and Boom! You have a beautiful portrait that captures all the little details of your pet, and always paints them in the best light! Huge thanks to @cubiq for his work on ComfyUI nodes and his awesome explainer series on youtube! Most of this pipeline was built and explained by him in his videos. If you want a pet portrait without going through all this trouble, consider buying one from here: pawprints.pinenlime.com! AI image manipulation tools have come a long way in the last few years. It’s quite easy nowadays to create beautiful, stylistic portraits of people and animals. But AI models are highly unpredictable. So most tools rely on the user (or some human) to weed out bad generations and find the best one. This is the classic “human in the loop” problem that often plagues AI tools. Turns out, with some clever tricks and careful tuning, you can build a pipeline that reliably works for the vast majority of pets. It is extremely resilient to variations in pose, lighting etc. In this post, I’ll dive deeper into how it works and all the neat little tricks that enable this. Here’s some examples of portraits you can generate with this pipeline. Let’s get started! The Key Ingredients IPAdapter The crux of the technique is the IPAdapter. It’s essentially a way to prompt a model using an image instead of text (it literally stands for Image Prompt Adapter). So instead of taking a text embeddings, it uses an image to get the embeddings. This is super powerful because it can accurately capture the style and structure in an image directly, instead of someone having to translate what they want from an image into text. In our ComfyUI IPAdapter node, we have two inputs, one for style and one for composition. We use a watercolor painting image for style, and feed the original image for composition (since we want to keep the same composition, but change the style). ControlNets Now that we have a way to keep the style consistent, we can turn our attention to faithfully representing the pet. IPAdapters bias heavily towards image quality, and image likeness suffers. So we need to do something to be able to keep the output looking like the same object as the input. The answer to that is ControlNet. ControlNets are another cool technique to provide additional constraints to the image generation process. Using a ControlNet, you can specify constraints in the form of edges, depth, human pose etc. A great feature of ControlNets is that they can be stacked. So you can have an edge controlnet that forces the output to have similar edges as the input, and also a depth controlnet that forces the output to have a similar depth profile. And that’s exactly what I do here. It turns out that controlnets are not just stackable with other controlnets, but they can also work in tandem with IPAdapter mentioned above. And so those are the tools we’ll use for this - IPAdapter with a source image to get the style, ControlNet with canny edge detector to constrain based on edges, and controlnet with depth to constrain based on depth profile. That’s really all you need in terms of techniques, but what I’ve learnt from experimenting with machine learning for production is that a lot of the value of these things comes from taking the time to tune all your parameters perfectly. So I want to talk a little bit about that. Finetuning Have you ever found a model with amazing example outputs, tried it on your own images and realized they look terrible? Often, the only reason for that is that the model hasn’t been finetuned for your images. It can feel like a complete blocker sometimes, because where do we even start with finetuning a pre-trained model?! Here’s what I’ve learnt about the topic. This extends beyond this particular pipeline too, so it’s good knowledge to have generally. ControlNets ControlNets are very powerful, so you have to be careful to control their effect on the output. Luckily, the custom nodes in ComfyUI let us reduce the effect of controlnets, and also stop their effects at any point. So we set the edge detector to 75% strength and make it stop influencing the generation at 75% and the depth detector stops at 30%. The reason we stop them at the end rather than just reducing their strength is that this allows the network to “clean up” any artifacts caused by them in the last few steps without being constrained externally. It just makes the image prettier. So it’s using just it’s training data to make things look as nice as possible, ignoring the edges and depth. The other big thing to tune is the KSampler. There’s a lot of little things going on here, but I’ll just briefly touch on some of them: KSampler - Steps First we have the steps. This is literally the number of times the model will repeatedly run. The more it runs, the more stylized your output gets, and the further away it will get from the original image. The effects of this are often not that obvious, so it’s worth playing around with it. KSampler - CFG Then there’s CFG. I honestly don’t fully understand this, but from its name - Classifier Free Guidance - I’m assuming it controls how much the model is allowed to modify the image unconstrained by the prompts in order to make it look better. This also affects the output image significantly, so is worth playing around with. Denoise Another neat little trick I use here is to start the image generation process with the input image  instead of a blank image, and keep the denoising low. This ensures that the output will look similar in terms of colors and textures. Text Prompt One thing you’ll notice I never mentioned is the text prompt until now! Surprising, since that’s usually the only conditioning that you normally provide to diffusion models. But in this case, we have so many other ways of conditioning that the text prompts usually just get in the way. So in this case, the prompt is literally just “a dog”. I do use the text prompt a little more in some of the more stylized portraits, like the chef dog or the one in the bathroom. Adding accessories In essence, this is more or less just an “AI Filter” that converts images into watercolor portraits. But it’s amazing how flexible this can be. For example, to make the portrait of the dog taking a shower, I literally just put images together in an image editing tool and used that as the input! The model takes care of unifying everything and cleaning up the image. Conclusion Now remove the background, add some text and Boom! You have a beautiful portrait that captures all the little details of your pet, and always paints them in the best light! Huge thanks to @cubiq for his work on ComfyUI nodes and his awesome explainer series on youtube! Most of this pipeline was built and explained by him in his videos. If you want a pet portrait without going through all this trouble, consider buying one from here: pawprints.pinenlime.com!

Make Masterpiece Pet Portraits with ComfyUI

ComfyUI এর সাথে মাস্টারপিস পোট্রেট পোট্রেট তৈরি করুন

Erstellen Sie meisterhafte Tierporträts mit ComfyUI

Crea retratos de mascotas de gran calidad con ComfyUI

Créez des portraits d'animaux de compagnie de qualité avec ComfyUI

ComfyUI के साथ पालतू जानवरों के बेहतरीन चित्र बनाएं

ComfyUI で傑作ペットポートレートを作ろう

ComfyUI로 걸작 애완동물 초상화 만들기

Crie retratos de animais de estimação de obras-primas com ComfyUI

Создавайте шедевральные портреты домашних животных с помощью ComfyUI

ComfyUI ile Başyapıt Evcil Hayvan Portreleri Yapın

Tạo ảnh chân dung thú cưng tuyệt đẹp với ComfyUI

使用 ComfyUI 制作宠物肖像杰作

Blocksquare has announced the launch of its new Marketplace Pool funding round. This development marks a significant milestone in real estate tokenization, leveraging the capabilities of its latest platform, Oceanpoint DeFi. The funding round, scheduled to commence on July 30th, introduces a novel approach to fractional investment in the hospitality sector. Through this initiative, Blocksquare and its partner, Pieme, aim to make hotel investments accessible to more investors. The global hotel industry, valued at over $700 billion in 2023, is experiencing rapid changes driven by technological advancements and shifting investment paradigms. As traditional investment avenues become saturated, fractional investment platforms like Blocksquare's Oceanpoint are emerging as viable alternatives. This shift is not just a trend but a response to increasing demand for more inclusive investment opportunities. Pieme’s Role in Democratizing Hotel Investment Pieme, a blockchain platform focused on democratizing hotel ownership, will be the marketplace manager on Blocksquare’s real estate tokenization infrastructure. The platform allows individuals to invest in hotel projects starting from as little as $150. This low entry point aims to break down barriers for smaller investors with limited access to high-value real estate ventures. The Pieme platform facilitates investment and provides a comprehensive management solution. Investors can monitor bookings, oversee stays, and engage with hotel operations via blockchain technology. Pieme handles operational, maintenance, and marketing tasks, streamlining the management process for investors. Pieme initially focuses on a hotel-residence project in Kampala, Uganda, with aspirations to expand across other emerging African markets. This strategic move aims to capitalize on the high growth potential within these regions, offering investors new opportunities in a burgeoning sector. Expanding into Africa reflects a broader trend of leveraging emerging markets for investment growth. According to the World Bank, the African real estate sector is expected to grow at a compound annual growth rate (CAGR) of 6.4% from 2024 to 2029. Pieme’s initiative is poised to attract attention from investors looking to diversify their portfolios and tap into these rapidly developing markets. The Oceanpoint Advantage Oceanpoint’s Marketplace Pools allow tokenized real estate ventures to be created and managed within the Blocksquare ecosystem. By submitting project proposals and gaining community support through a democratic voting process, marketplace operators can secure funding from staked BST tokens (sBST). This system fosters a community-driven approach to real estate investment, enhancing transparency and engagement. The success of the previous Oceanpoint Launchpad campaign, which sold out within 36 hours, underscores the growing interest in tokenized real estate investments. This rapid sell-out highlights the market’s readiness for innovative investment solutions and the potential for significant returns. Upcoming Engagement For those interested in learning more about the Pieme Marketplace Pool Campaign, Blocksquare and Pieme will host a Block-chat on Thursday, July 25th at 6:00 PM UTC. This session will provide further insights into the funding round and offer an opportunity for prospective investors to engage with the project. Blocksquare’s foray into fractional hotel investments through Pieme represents a significant step in the evolution of real estate investment. As the landscape continues to shift, platforms like Oceanpoint and Pieme are set to play a pivotal role in shaping the future of hospitality investments. Don’t forget to like and share the story! Vested Interest Disclosure: This author is an independent contributor publishing via our business blogging program. HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYOR. Blocksquare has announced the launch of its new Marketplace Pool funding round. This development marks a significant milestone in real estate tokenization, leveraging the capabilities of its latest platform, Oceanpoint DeFi. The funding round, scheduled to commence on July 30th, introduces a novel approach to fractional investment in the hospitality sector. Through this initiative, Blocksquare and its partner, Pieme, aim to make hotel investments accessible to more investors. The global hotel industry, valued at over $700 billion in 2023, is experiencing rapid changes driven by technological advancements and shifting investment paradigms. As traditional investment avenues become saturated, fractional investment platforms like Blocksquare's Oceanpoint are emerging as viable alternatives. This shift is not just a trend but a response to increasing demand for more inclusive investment opportunities. Pieme’s Role in Democratizing Hotel Investment Pieme, a blockchain platform focused on democratizing hotel ownership, will be the marketplace manager on Blocksquare’s real estate tokenization infrastructure. The platform allows individuals to invest in hotel projects starting from as little as $150. This low entry point aims to break down barriers for smaller investors with limited access to high-value real estate ventures. The Pieme platform facilitates investment and provides a comprehensive management solution. Investors can monitor bookings, oversee stays, and engage with hotel operations via blockchain technology. Pieme handles operational, maintenance, and marketing tasks, streamlining the management process for investors. Pieme initially focuses on a hotel-residence project in Kampala, Uganda, with aspirations to expand across other emerging African markets. This strategic move aims to capitalize on the high growth potential within these regions, offering investors new opportunities in a burgeoning sector. Expanding into Africa reflects a broader trend of leveraging emerging markets for investment growth. According to the World Bank, the African real estate sector is expected to grow at a compound annual growth rate (CAGR) of 6.4% from 2024 to 2029. Pieme’s initiative is poised to attract attention from investors looking to diversify their portfolios and tap into these rapidly developing markets. The Oceanpoint Advantage Oceanpoint’s Marketplace Pools allow tokenized real estate ventures to be created and managed within the Blocksquare ecosystem. By submitting project proposals and gaining community support through a democratic voting process, marketplace operators can secure funding from staked BST tokens (sBST). This system fosters a community-driven approach to real estate investment, enhancing transparency and engagement. The success of the previous Oceanpoint Launchpad campaign, which sold out within 36 hours, underscores the growing interest in tokenized real estate investments. This rapid sell-out highlights the market’s readiness for innovative investment solutions and the potential for significant returns. Upcoming Engagement For those interested in learning more about the Pieme Marketplace Pool Campaign, Blocksquare and Pieme will host a Block-chat on Thursday, July 25th at 6:00 PM UTC. This session will provide further insights into the funding round and offer an opportunity for prospective investors to engage with the project. Blocksquare’s foray into fractional hotel investments through Pieme represents a significant step in the evolution of real estate investment. As the landscape continues to shift, platforms like Oceanpoint and Pieme are set to play a pivotal role in shaping the future of hospitality investments. Blocksquare’s Don’t forget to like and share the story! Vested Interest Disclosure: This author is an independent contributor publishing via our business blogging program. HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYOR. Vested Interest Disclosure: This author is an independent contributor publishing via our business blogging program . HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYOR. Vested Interest Disclosure: business blogging program business blogging program

Blocksquare Brings Tokenized Hotel Ownership to Retail; Launches Pooling Marketplace

ব্লকস্কয়ার টোকেনাইজড হোটেল মালিকানা নিয়ে আসে খুচরা বিক্রেতা; পুলিং মার্কেটপ্লেস চালু করে

Blocksquare bringt tokenisierten Hotelbesitz in den Einzelhandel und startet Pooling-Marktplatz

Blocksquare lleva la propiedad hotelera tokenizada al comercio minorista; Lanza el mercado de pooling

Blocksquare apporte la propriété hôtelière tokenisée au commerce de détail ; Lance le marché de pooling

ब्लॉकस्क्वेयर टोकनयुक्त होटल स्वामित्व को रिटेल में लाता है; पूलिंग मार्केटप्लेस लॉन्च करता है

Blocksquare がトークン化されたホテル所有権を小売に導入、プーリング マーケットプレイスを開始

Blocksquare는 토큰화된 호텔 소유권을 소매업에 도입합니다. 풀링 마켓플레이스 출시

Blocksquare traz propriedade hoteleira tokenizada para o varejo; Lança mercado de pooling

Blocksquare предлагает токенизированное владение отелями в розничной торговле; Запускает торговую площадку пула

Blocksquare Tokenize Otel Sahipliğini Perakendeye Getiriyor; Havuz Pazarını Başlattı

Blocksquare mang quyền sở hữu khách sạn được mã hóa sang lĩnh vực bán lẻ; Ra mắt thị trường Pooling

Blocksquare 将代币化酒店所有权引入零售业；推出共享市场

How a Herd of Models Challenges ChatGPT's Dominance: Conclusion, Discussion, and References

Authors:
(1) Zengyi Qin, MIT & MyShell.ai and (email: qinzy@mit.edu);
(2) Wenliang Zhao, Tsinghua University;
(3) Xumin Yu, Tsinghua University;
(4) Xin Sun, MyShell.ai; Table of Links Abstract and Introduction ApproachExperiment Discussion and References 4 Discussion OpenVoice demonstrates remarkable instance voice cloning capabilities and is more flexible than previous approaches in terms of voice styles and languages. The intuition behind the approach is that it is relatively easy to train a base speaker TTS model to control the voice styles and languages, as long as we do not require the model to have the ability to clone the tone color of the reference speaker. Therefore, we proposed to decouple the tone color cloning from the remaining voice styles and the language, which we believe is the foundational design principle of OpenVoice. In order to facilitate future research, we make the source code and model weights publicly available. References [1] I. P. Association. Handbook of the International Phonetic Association: A guide to the use of the International Phonetic Alphabet. Cambridge University Press, 1999. [2] E. Casanova, J. Weber, C. D. Shulby, A. C. Junior, E. Gölge, and M. A. Ponti. Yourtts: Towards zero-shot multi-speaker tts and zero-shot voice conversion for everyone. In International Conference on Machine Learning, pages 2709–2720. PMLR, 2022. [3] CoquiAI. Xtts taking text-to-speech to the next level. Technical Blog, 2023. [4] W.-N. Hsu, B. Bolte, Y.-H. H. Tsai, K. Lakhotia, R. Salakhutdinov, and A. Mohamed. Hubert: Self-supervised speech representation learning by masked prediction of hidden units. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29:3451–3460, 2021. [5] J. Kim, S. Kim, J. Kong, and S. Yoon. Glow-tts: A generative flow for text-to-speech via monotonic alignment search. Advances in Neural Information Processing Systems, 33:8067– 8077, 2020. [6] J. Kim, J. Kong, and J. Son. Conditional variational autoencoder with adversarial learning for end-to-end text-to-speech. In International Conference on Machine Learning, pages 5530–5540. PMLR, 2021. [7] J. Kong, J. Kim, and J. Bae. Hifi-gan: Generative adversarial networks for efficient and high fidelity speech synthesis. Advances in Neural Information Processing Systems, 33:17022–17033, 2020. [8] M. Le, A. Vyas, B. Shi, B. Karrer, L. Sari, R. Moritz, M. Williamson, V. Manohar, Y. Adi, J. Mahadeokar, et al. Voicebox: Text-guided multilingual universal speech generation at scale. arXiv preprint arXiv:2306.15687, 2023. [9] J. Li, W. Tu, and L. Xiao. Freevc: Towards high-quality text-free one-shot voice conversion. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5. IEEE, 2023. [10] M. Müller. Dynamic time warping. Information retrieval for music and motion, pages 69–84, 2007. [11] A. Polyak, Y. Adi, J. Copet, E. Kharitonov, K. Lakhotia, W.-N. Hsu, A. Mohamed, and E. Dupoux. Speech resynthesis from discrete disentangled self-supervised representations. arXiv preprint arXiv:2104.00355, 2021. [12] D. Rezende and S. Mohamed. Variational inference with normalizing flows. In International conference on machine learning, pages 1530–1538. PMLR, 2015. [13] P. Senin. Dynamic time warping algorithm review. Information and Computer Science Department University of Hawaii at Manoa Honolulu, USA, 855(1-23):40, 2008. [14] B. van Niekerk, M.-A. Carbonneau, J. Zaïdi, M. Baas, H. Seuté, and H. Kamper. A comparison of discrete and soft speech units for improved voice conversion. In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 6562– 6566. IEEE, 2022. [15] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017. [16] C. Wang, S. Chen, Y. Wu, Z. Zhang, L. Zhou, S. Liu, Z. Chen, Y. Liu, H. Wang, J. Li, et al. Neural codec language models are zero-shot text to speech synthesizers. arXiv preprint arXiv:2301.02111, 2023. [17] D. Yang, S. Liu, R. Huang, G. Lei, C. Weng, H. Meng, and D. Yu. Instructtts: Modelling expressive tts in discrete latent space with natural language style prompt. arXiv preprint arXiv:2301.13662, 2023. [18] Z. Zhang, L. Zhou, C. Wang, S. Chen, Y. Wu, S. Liu, Z. Chen, Y. Liu, H. Wang, J. Li, et al. Speak foreign languages with your own voice: Cross-lingual neural codec language modeling. arXiv preprint arXiv:2303.03926, 2023. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Zengyi Qin, MIT & MyShell.ai and (email: qinzy@mit.edu); (2) Wenliang Zhao, Tsinghua University; (3) Xumin Yu, Tsinghua University; (4) Xin Sun, MyShell.ai; Authors: Authors: (1) Zengyi Qin, MIT & MyShell.ai and (email: qinzy@mit.edu); (2) Wenliang Zhao, Tsinghua University; (3) Xumin Yu, Tsinghua University; (4) Xin Sun, MyShell.ai; Table of Links Abstract and Introduction Abstract and Introduction Approach Experiment Approach Experiment Discussion and References Discussion and References 4 Discussion OpenVoice demonstrates remarkable instance voice cloning capabilities and is more flexible than previous approaches in terms of voice styles and languages. The intuition behind the approach is that it is relatively easy to train a base speaker TTS model to control the voice styles and languages, as long as we do not require the model to have the ability to clone the tone color of the reference speaker. Therefore, we proposed to decouple the tone color cloning from the remaining voice styles and the language, which we believe is the foundational design principle of OpenVoice. In order to facilitate future research, we make the source code and model weights publicly available. References [1] I. P. Association. Handbook of the International Phonetic Association: A guide to the use of the International Phonetic Alphabet. Cambridge University Press, 1999. [2] E. Casanova, J. Weber, C. D. Shulby, A. C. Junior, E. Gölge, and M. A. Ponti. Yourtts: Towards zero-shot multi-speaker tts and zero-shot voice conversion for everyone. In International Conference on Machine Learning, pages 2709–2720. PMLR, 2022. [3] CoquiAI. Xtts taking text-to-speech to the next level. Technical Blog, 2023. [4] W.-N. Hsu, B. Bolte, Y.-H. H. Tsai, K. Lakhotia, R. Salakhutdinov, and A. Mohamed. Hubert: Self-supervised speech representation learning by masked prediction of hidden units. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29:3451–3460, 2021. [5] J. Kim, S. Kim, J. Kong, and S. Yoon. Glow-tts: A generative flow for text-to-speech via monotonic alignment search. Advances in Neural Information Processing Systems, 33:8067– 8077, 2020. [6] J. Kim, J. Kong, and J. Son. Conditional variational autoencoder with adversarial learning for end-to-end text-to-speech. In International Conference on Machine Learning, pages 5530–5540. PMLR, 2021. [7] J. Kong, J. Kim, and J. Bae. Hifi-gan: Generative adversarial networks for efficient and high fidelity speech synthesis. Advances in Neural Information Processing Systems, 33:17022–17033, 2020. [8] M. Le, A. Vyas, B. Shi, B. Karrer, L. Sari, R. Moritz, M. Williamson, V. Manohar, Y. Adi, J. Mahadeokar, et al. Voicebox: Text-guided multilingual universal speech generation at scale. arXiv preprint arXiv:2306.15687, 2023. [9] J. Li, W. Tu, and L. Xiao. Freevc: Towards high-quality text-free one-shot voice conversion. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5. IEEE, 2023. [10] M. Müller. Dynamic time warping. Information retrieval for music and motion, pages 69–84, 2007. [11] A. Polyak, Y. Adi, J. Copet, E. Kharitonov, K. Lakhotia, W.-N. Hsu, A. Mohamed, and E. Dupoux. Speech resynthesis from discrete disentangled self-supervised representations. arXiv preprint arXiv:2104.00355, 2021. [12] D. Rezende and S. Mohamed. Variational inference with normalizing flows. In International conference on machine learning, pages 1530–1538. PMLR, 2015. [13] P. Senin. Dynamic time warping algorithm review. Information and Computer Science Department University of Hawaii at Manoa Honolulu, USA, 855(1-23):40, 2008. [14] B. van Niekerk, M.-A. Carbonneau, J. Zaïdi, M. Baas, H. Seuté, and H. Kamper. A comparison of discrete and soft speech units for improved voice conversion. In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 6562– 6566. IEEE, 2022. [15] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017. [16] C. Wang, S. Chen, Y. Wu, Z. Zhang, L. Zhou, S. Liu, Z. Chen, Y. Liu, H. Wang, J. Li, et al. Neural codec language models are zero-shot text to speech synthesizers. arXiv preprint arXiv:2301.02111, 2023. [17] D. Yang, S. Liu, R. Huang, G. Lei, C. Weng, H. Meng, and D. Yu. Instructtts: Modelling expressive tts in discrete latent space with natural language style prompt. arXiv preprint arXiv:2301.13662, 2023. [18] Z. Zhang, L. Zhou, C. Wang, S. Chen, Y. Wu, S. Liu, Z. Chen, Y. Liu, H. Wang, J. Li, et al. Speak foreign languages with your own voice: Cross-lingual neural codec language modeling. arXiv preprint arXiv:2303.03926, 2023. 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

OpenVoice: Versatile Instant Voice Cloning - Discussion and References

Enabling Acoustic Audience Feedback in Large Virtual Events: Conclusions & References

This paper is available on arxiv under CC 4.0 license. Authors: (1) Muzhaffar Hazman, University of Galway, Ireland; (2) Susan McKeever, Technological University Dublin, Ireland; (3) Josephine Griffith, University of Galway, Ireland. Table of Links Abstract and Introduction Related Works Methodology Results Limitations and Future Works Conclusion, Acknowledgments, and References A Hyperparameters and Settings B Metric: Weighted F1-Score C Architectural Details D Performance Benchmarking E Contingency Table: Baseline vs. Text-STILT 6 Conclusion In this work, we addressed the challenge of training multimodal meme sentiment classifiers on a limited number of labelled memes by incorporating unimodal sentiment analysis data. We did so by proposing the first instance of STILT that applies unimodal intermediate tasks to a multimodal target task. Specifically, we tested image-only and text-only sentiment classification as intermediate tasks in training a meme sentiment classifier. We showed that this approach worked – unimodal text improved meme classification performance to a statistically significant degree. This novel approach allowed us to train a meme classifier that outperforms meme-only finetuning with only 60% as many labelled meme samples. As possible explanations for our observations, we discuss apparent similarities and differences in the roles of image and text modalities between unimodal and multimodal sentiment analysis tasks. Acknowledgments This work was conducted with the financial support of the Science Foundation Ireland Centre for Research Training in Digitally-Enhanced Reality (d-real) under Grant No. 18/CRT/6224; and the provision of computational facilities and support from the Irish Centre for High-End Computing (ICHEC). References Christian Bauckhage. 2011. Insights into internet memes. Proceedings of the International AAAI Conference on Web and Social Media, 5(1):42–49. Efrat Blaier, Itzik Malkiel, and Lior Wolf. 2021. Caption enriched samples for improving hateful memes detection. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 9350–9358, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics. Ana-Maria Bucur, Adrian Cosma, and Ioana Iordache. 2022. BLUE at memotion 2.0 2022: You have my image, my text and my transformer. In De-Factify @ AAAI 2022. First Workshop on Multimodal FactChecking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI. Christopher Clark, Kenton Lee, Ming-Wei Chang, Tom Kwiatkowski, Michael Collins, and Kristina Toutanova. 2019. BoolQ: Exploring the surprising difficulty of natural yes/no questions. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 2924–2936, Minneapolis, Minnesota. Association for Computational Linguistics. CrowdFlower. 2016. Image sentiment polarity - dataset. Available at https://data.world/ crowdflower/image-sentiment-polarity. Accessed: 2023-01-15. Baishan Duan and Yuesheng Zhu. 2022. Browallia at memotion 2.0 2022 : Multimodal memotion analysis with modified ogb strategies (short paper). In DeFactify @ AAAI 2022. First Workshop on Multimodal Fact-Checking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI. Noam Gal, Limor Shifman, and Zohar Kampf. 2016. “it gets better”: Internet memes and the construction of collective identity. New Media & Society, 18(8):1698–1714. Yue Gu, Kangning Yang, Shiyu Fu, Shuhong Chen, Xinyu Li, and Ivan Marsic. 2018. Hybrid Attention based Multimodal Network for Spoken Language Classification. In Proceedings of the 27th International Conference on Computational Linguistics, pages 2379–2390, Santa Fe, New Mexico, USA. Association for Computational Linguistics. Muzhaffar Hazman, Susan McKeever, and Josephine Griffith. 2023. Meme sentiment analysis enhanced with multimodal spatial encoding and face embedding. In Artificial Intelligence and Cognitive Science, pages 318–331, Cham. Springer Nature Switzerland. Junguang Jiang, Yang Shu, Jianmin Wang, and Mingsheng Long. 2022. Transferability in deep learning: A survey. arXiv preprint, arXiv:2201.05867. Vishal Keswani, Sakshi Singh, Suryansh Agarwal, and Ashutosh Modi. 2020. IITK at SemEval-2020 task 8: Unimodal and bimodal sentiment analysis of Internet memes. In Proceedings of the Fourteenth Workshop on Semantic Evaluation, pages 1135–1140, Barcelona (online). International Committee for Computational Linguistics. Douwe Kiela, Hamed Firooz, Aravind Mohan, Vedanuj Goswami, Amanpreet Singh, Pratik Ringshia, et al. 2020. The hateful memes challenge: Detecting hate speech in multimodal memes. Advances in Neural Information Processing Systems, 33:2611–2624. Liisi Laineste and Piret Voolaid. 2017. Laughing across borders: Intertextuality of internet memes. The European Journal of Humour Research, 4(4):26–49. Gwang Gook Lee and Mingwei Shen. 2022. Amazon pars at memotion 2.0 2022: Multi-modal multi-task learning for memotion 2.0 challenge. In De-Factify @ AAAI 2022. First Workshop on Multimodal FactChecking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI. Quinn McNemar. 1947. Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika, 12(2):153–157. Shreyash Mishra, S Suryavardan, Parth Patwa, Megha Chakraborty, Anku Rani, Aishwarya Reganti, Aman Chadha, Amitava Das, Amit Sheth, Manoj Chinnakotla, Asif Ekbal, and Srijan Kumar. 2023. Memotion 3: Dataset on sentiment and emotion analysis of codemixed hindi-english memes. arXiv preprint, arXiv:2303.09892. Niklas Muennighoff. 2020. Vilio: state-of-the-art visiolinguistic models applied to hateful memes. arXiv preprint arXiv:2012.07788. Thanh Van Nguyen, Nhat Truong Pham, Ngoc Duy Nguyen, Hai Nguyen, Long H. Nguyen, and YongGuk Kim. 2022. Hcilab at memotion 2.0 2022: Analysis of sentiment, emotion and intensity of emotion classes from meme images using single and multi modalities (short paper). In De-Factify @ AAAI 2022. First Workshop on Multimodal Fact-Checking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI. Asaf Nissenbaum and Limor Shifman. 2017. Internet memes as contested cultural capital: The case of 4chan’s /b/ board. New Media & Society, 19(4):483– 501. Parth Patwa, Sathyanarayanan Ramamoorthy, Nethra Gunti, Shreyash Mishra, S Suryavardan, Aishwarya Reganti, et al. 2022. Findings of memotion 2: Sentiment and emotion analysis of memes. In De-Factify @ AAAI 2022. First Workshop on Multimodal FactChecking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI. Kim Ngan Phan, Gueesang Lee, Hyung-Jeong Yang, and Soo-Hyung Kim. 2022. Little flower at memotion 2.0 2022 : Ensemble of multi-modal model using attention mechanism in memotion analysis (short paper). In De-Factify @ AAAI 2022. First Workshop on Multimodal Fact-Checking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI. Jason Phang, Thibault Fevry, and Samuel R. Bowman. ´ 2019. Sentence encoders on stilts: Supplementary training on intermediate labeled-data tasks. arXiv preprint, arXiv:1811.01088. Mohammad Taher Pilehvar and Jose Camacho-Collados. 2019. WiC: the word-in-context dataset for evaluating context-sensitive meaning representations. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 1267–1273, Minneapolis, Minnesota. Association for Computational Linguistics. Clifton Poth, Jonas Pfeiffer, Andreas Ruckl ¨ e, and Iryna ´ Gurevych. 2021. What to pre-train on? Efficient intermediate task selection. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 10585–10605, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics. Christopher Potts, Zhengxuan Wu, Atticus Geiger, and Douwe Kiela. 2021. DynaSent: A dynamic benchmark for sentiment analysis. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 2388–2404, Online. Association for Computational Linguistics. Shraman Pramanick, Md Shad Akhtar, and Tanmoy Chakraborty. 2021a. Exercise? i thought you said ’extra fries’: Leveraging sentence demarcations and multi-hop attention for meme affect analysis. Proceedings of the International AAAI Conference on Web and Social Media, 15(1):513–524. Shraman Pramanick, Shivam Sharma, Dimitar Dimitrov, Md. Shad Akhtar, Preslav Nakov, and Tanmoy Chakraborty. 2021b. MOMENTA: A Multimodal Framework for Detecting Harmful Memes and Their Targets. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 4439–4455, Punta Cana, Dominican Republic. Association for Computational Linguistics. Yada Pruksachatkun, Jason Phang, Haokun Liu, Phu Mon Htut, Xiaoyi Zhang, Richard Yuanzhe Pang, Clara Vania, Katharina Kann, and Samuel R. Bowman. 2020. Intermediate-task transfer learning with pretrained language models: When and why does it work? In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5231–5247, Online. Association for Computational Linguistics. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, et al. 2021. Learning transferable visual models from natural language supervision. In International Conference on Machine Learning, pages 8748–8763. PMLR. Sathyanarayanan Ramamoorthy, Nethra Gunti, Shreyash Mishra1, S Suryavardan, Aishwarya Reganti, Parth Patwa, Amitava Das, Tanmoy Chakraborty, Amit Sheth, Asif Ekbal, and Chaitanya Ahuja. 2022. Memotion 2: Dataset on sentiment and emotion analysis of memes. In De-Factify @ AAAI 2022. First Workshop on Multimodal Fact-Checking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI. Elad Segev, Asaf Nissenbaum, Nathan Stolero, and Limor Shifman. 2015. Families and Networks of Internet Memes: The Relationship Between Cohesiveness, Uniqueness, and Quiddity Concreteness. Journal of Computer-Mediated Communication, 20(4):417–433. Lanyu Shang, Yang Zhang, Yuheng Zha, Yingxi Chen, Christina Youn, and Dong Wang. 2021. AOMD: An Analogy-Aware Approach to Offensive Meme Detection on Social Media. Inf. Process. Manage., 58(5). Chhavi Sharma, Deepesh Bhageria, William Scott, Srinivas PYKL, Amitava Das, Tanmoy Chakraborty, et al. 2020. SemEval-2020 task 8: Memotion analysisthe visuo-lingual metaphor! In Proceedings of the Fourteenth Workshop on Semantic Evaluation, pages 759–773, Barcelona (online). International Committee for Computational Linguistics. Shivam Sharma, Mohd Khizir Siddiqui, Md. Shad Akhtar, and Tanmoy Chakraborty. 2022. Domainaware self-supervised pre-training for label-efficient meme analysis. In Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 792–805, Online only. Association for Computational Linguistics. Limor Shifman. 2014. The cultural logic of photo-based meme genres. Journal of Visual Culture, 13(3):340–358. Pranaydeep Singh, Aaron Maladry, and Els Lefever. 2022. Combining language models and linguistic information to label entities in memes. In Proceedings of the Workshop on Combating Online Hostile Posts in Regional Languages during Emergency Situations, pages 35–42, Dublin, Ireland. Association for Computational Linguistics. Shardul Suryawanshi, Bharathi Raja Chakravarthi, Pranav Verma, Mihael Arcan, John Philip McCrae, and Paul Buitelaar. 2020. A dataset for troll classification of TamilMemes. In Proceedings of the WILDRE5– 5th Workshop on Indian Language Data: Resources and Evaluation, pages 7–13, Marseille, France. European Language Resources Association (ELRA). Cagri Toraman, Furkan S¸ahinuc¸, and Eyup Yilmaz. 2022. Large-scale hate speech detection with crossdomain transfer. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 2215–2225, Marseille, France. European Language Resources Association. Alex Wang, Jan Hula, Patrick Xia, Raghavendra Pappagari, R. Thomas McCoy, Roma Patel, Najoung Kim, Ian Tenney, Yinghui Huang, Katherin Yu, Shuning Jin, Berlin Chen, Benjamin Van Durme, Edouard Grave, Ellie Pavlick, and Samuel R. Bowman. 2019. Can you tell me how to get past sesame street? sentence-level pretraining beyond language modeling. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4465–4476, Florence, Italy. Association for Computational Linguistics. Ron Zhu. 2020. Enhance multimodal transformer with external label and in-domain pretrain: Hateful meme challenge winning solution. arXiv preprint, arXiv:2012.08290. Yan Zhuang and Yanru Zhang. 2022. Yet at memotion 2.0 2022 : Hate speech detection combining bilstm and fully connected layers (short paper). In De-Factify @ AAAI 2022. First Workshop on Multimodal Fact-Checking and Hate Speech Detection, CEUR Workshop Proceedings. AAAI.

Innovative Strategies in Meme Sentiment Analysis: The Impact of Unimodal Training

Stories, News & Mentions about KLAMP

Klamp’s Stories on HackerNoon