Nurall

nurall

Lifelong Intelligence Beyond the Edge using Hyperdimensional Computing: Conclusion, and References

Authors:
(1) Xiaofan Yu, University of California San Diego, La Jolla, California, USA (x1yu@ucsd.edu);
(2) Anthony Thomas, University of California San Diego, La Jolla, California, USA (ahthomas@ucsd.edu);
(3) Ivannia Gomez Moreno, CETYS University, Campus Tijuana, Tijuana, Mexico (ivannia.gomez@cetys.edu.mx);
(4) Louis Gutierrez, University of California San Diego, La Jolla, California, USA (l8gutierrez@ucsd.edu);
(5) Tajana Šimunić Rosing, University of California San Diego, La Jolla, USA (tajana@ucsd.edu). Table of Links Abstract and 1. Introduction 2 Related Work 3 Background on HDC 4 Problem Definition 5 LifeDH 6 Variants of LifeHD 7 Evaluation of LifeHD 8 Evaluation of LifeHD semi and LifeHDa 9 Discussions and Future Works 10 Conclusion, Acknowledgments, and References 2 RELATED WORK Lifelong and On-Device Learning. Lifelong learning (or continual learning) is a large and active area of research in the broader machine learning community. Catastrophic forgetting is a major challenge in lifelong learning, and refers to a commonly observed empirical phenomenon in which updating certain machine learning models with new data severely degrades their ability to perform previously learned tasks [36]. Previous works proposed techniques such as dynamic architecture [31, 49], regularization by penalizing important weights [28, 66], knowledge distillation from past models [14] and experience replay using a memory buffer [35, 58]. The lifelong learning literature has examined a wide range of problem settings, ranging from the fully supervised case, in which tasks and class labels are provided, and the fully unsupervised case without any labels and prior knowledge [13, 57]. However, all of these works are based on deep NNs and require backpropagation, which is problematic for resource-constrained devices. Neurally-inspired lightweight algorithms have recently been proposed for lifelong learning applications. FlyModel [52] and SDMLP [8] use sparse coding and associative memory for lifelong learning. However, both approaches assume full supervision. STAM [54] is an expandable memory architecture for unsupervised lifelong learning, using layered receptive fields and a two-tier memory hierarchy. It learns via online centroid-based clustering pipeline, novelty detection and memory updates. Nevertheless, the memory in STAM is solely dedicated to image storage, while our LifeHD additionally emphasizes merging past patterns into coarse groups and shows more effective learning performance. Recent works optimize the resource usage of on-device training via pruning and quantization [34, 45], tuning partial weights [9, 47], memory profiling and optimization [15, 61, 64], as well as growing the NN on the fly [67]. All these works optimize training given resource constraints and do not focus on lifelong learning. They are orthogonal to the contribution of LifeHD which focuses on adaptive and continual training. LifeHD can be further optimized by combining with such techniques. Hyperdimensional Computing. HDC has garnered substantial interest from the computer hardware community as an energyefficient and low-latency approach to learning, and has been successfully applied to problems such as human activity recognition [27], voice recognition [23], image recognition [11, 65], to name a few. The large majority of literature on HDC has focused on using the technique to perform supervised classification tasks. Among the limited literature for weakly-supervised learning with HDC, HDCluster [21] enabled unsupervised clustering in HDC with a new algorithm that is similar to K-Means. SemiHD [22] is a semi-supervised learning framework using HDC with iterative self-labeling. Hyperseed [41], C-FSCIL [18] and FSL-HD [65] adopted HDC or similar vector symbolic architectures (VSA) for unsupervised or few-shot learning. All above works did not consider the lifelong aspect and used offline training on a static dataset. To the best of the authors’ knowledge, LifeHD is the first work that designs and deploys lifelong learning in edge IoT applications especially with zero or minimal amount of labels. This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license. Authors: (1) Xiaofan Yu, University of California San Diego, La Jolla, California, USA (x1yu@ucsd.edu); (2) Anthony Thomas, University of California San Diego, La Jolla, California, USA (ahthomas@ucsd.edu); (3) Ivannia Gomez Moreno, CETYS University, Campus Tijuana, Tijuana, Mexico (ivannia.gomez@cetys.edu.mx); (4) Louis Gutierrez, University of California San Diego, La Jolla, California, USA (l8gutierrez@ucsd.edu); (5) Tajana Šimunić Rosing, University of California San Diego, La Jolla, USA (tajana@ucsd.edu). Authors: Authors: (1) Xiaofan Yu, University of California San Diego, La Jolla, California, USA (x1yu@ucsd.edu); (2) Anthony Thomas, University of California San Diego, La Jolla, California, USA (ahthomas@ucsd.edu); (3) Ivannia Gomez Moreno, CETYS University, Campus Tijuana, Tijuana, Mexico (ivannia.gomez@cetys.edu.mx); (4) Louis Gutierrez, University of California San Diego, La Jolla, California, USA (l8gutierrez@ucsd.edu); (5) Tajana Šimunić Rosing, University of California San Diego, La Jolla, USA (tajana@ucsd.edu). Table of Links Abstract and 1. Introduction Abstract and 1. Introduction 2 Related Work 2 Related Work 3 Background on HDC 3 Background on HDC 4 Problem Definition 4 Problem Definition 5 LifeDH 5 LifeDH 6 Variants of LifeHD 6 Variants of LifeHD 7 Evaluation of LifeHD 7 Evaluation of LifeHD 8 Evaluation of LifeHD semi and LifeHDa 8 Evaluation of LifeHD semi and LifeHDa 9 Discussions and Future Works 9 Discussions and Future Works 10 Conclusion, Acknowledgments, and References 10 Conclusion, Acknowledgments, and References 2 RELATED WORK 2 RELATED WORK Lifelong and On-Device Learning. Lifelong learning (or continual learning) is a large and active area of research in the broader machine learning community. Catastrophic forgetting is a major challenge in lifelong learning, and refers to a commonly observed empirical phenomenon in which updating certain machine learning models with new data severely degrades their ability to perform previously learned tasks [36]. Previous works proposed techniques such as dynamic architecture [31, 49], regularization by penalizing important weights [28, 66], knowledge distillation from past models [14] and experience replay using a memory buffer [35, 58]. The lifelong learning literature has examined a wide range of problem settings, ranging from the fully supervised case, in which tasks and class labels are provided, and the fully unsupervised case without any labels and prior knowledge [13, 57]. However, all of these works are based on deep NNs and require backpropagation, which is problematic for resource-constrained devices. Lifelong and On-Device Learning. Neurally-inspired lightweight algorithms have recently been proposed for lifelong learning applications. FlyModel [52] and SDMLP [8] use sparse coding and associative memory for lifelong learning. However, both approaches assume full supervision. STAM [54] is an expandable memory architecture for unsupervised lifelong learning, using layered receptive fields and a two-tier memory hierarchy. It learns via online centroid-based clustering pipeline, novelty detection and memory updates. Nevertheless, the memory in STAM is solely dedicated to image storage, while our LifeHD additionally emphasizes merging past patterns into coarse groups and shows more effective learning performance. Recent works optimize the resource usage of on-device training via pruning and quantization [34, 45], tuning partial weights [9, 47], memory profiling and optimization [15, 61, 64], as well as growing the NN on the fly [67]. All these works optimize training given resource constraints and do not focus on lifelong learning. They are orthogonal to the contribution of LifeHD which focuses on adaptive and continual training. LifeHD can be further optimized by combining with such techniques. Hyperdimensional Computing. HDC has garnered substantial interest from the computer hardware community as an energyefficient and low-latency approach to learning, and has been successfully applied to problems such as human activity recognition [27], voice recognition [23], image recognition [11, 65], to name a few. The large majority of literature on HDC has focused on using the technique to perform supervised classification tasks. Among the limited literature for weakly-supervised learning with HDC, HDCluster [21] enabled unsupervised clustering in HDC with a new algorithm that is similar to K-Means. SemiHD [22] is a semi-supervised learning framework using HDC with iterative self-labeling. Hyperseed [41], C-FSCIL [18] and FSL-HD [65] adopted HDC or similar vector symbolic architectures (VSA) for unsupervised or few-shot learning. All above works did not consider the lifelong aspect and used offline training on a static dataset. To the best of the authors’ knowledge, LifeHD is the first work that designs and deploys lifelong learning in edge IoT applications especially with zero or minimal amount of labels. Hyperdimensional Computing. 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

Lifelong Intelligence Beyond the Edge using Hyperdimensional Computing: Related Work

Lifelong Intelligence Beyond the Edge using Hyperdimensional Computing: Abstract and Introduction

This paper is available on arxiv under CC0 1.0 DEED license. Authors: (1) Salim Chouaki, LIX, CNRS, Inria, Ecole Polytechnique, Institut Polytechnique de Paris; (2) Oana Goga, LIX, CNRS, Inria, Ecole Polytechnique, Institut Polytechnique de Paris; (3) Hamed Haddadi, Imperial College London, Brave Software; (4) Peter Snyder, Brave Software. Table of Links Abstract and Introduction Background Measurement Methodology Results Limitations Related Work Conclusion, Acknowledgments, and References Appendix 7 CONCLUSION In this paper, we presented the first systematic study of the privacy properties of the advertising systems of five popular search engines: Two traditional ones, Google and Bing, and three private ones, DuckDuckGo, StartPage, and Qwant. We investigated whether, and to which extent, search engines through their advertising systems, engage in privacyharming behaviors that allow cross site tracking. Despite the privacy intentions and promises of private search engines, our findings reveal the failure of privacyfocused search engines to fully protect users’ privacy during ad interactions. Users on all measured search engines, including the privacy-focused ones, are subject to navigation-based tracking by third parties. We find that all search engines engage in bounce tracking when clicking on ads, where users are sent through several redirectors before reaching the ads’ destination websites. While private search engines themselves do not engage in user tracking, their reliance on traditional advertising systems (Microsoft or Google) renders users susceptible to tracking by those systems. Although we cannot directly attribute this tracking to the search engines themselves, it is evident that they are enabling it through their reliance on Microsoft and Google’s advertising systems. Inspecting the privacy policies of the search engines in light of our findings reveals interesting disparities. While our results demonstrate that Microsoft is capable of tracking DuckDuckGo users when they click on ads, DuckDuckGo asserts that Microsoft does not associate ad-click data with user profiles. On the other hand, Qwant, which also relies on Microsoft advertising for a significant fraction of its ads, do not document the utilization of ad-click data by Microsoft and whether it is used to enhance user profiles. Similarly, StartPage explicitly states that clicking on ads subjects users to the data collection policies of other websites. Our study highlights the need for increased attention to privacy protection within the advertising systems of search engines. One potential solution to protect users’ privacy for private search engines would be to reduce their reliance on third-party advertising systems. Developing their own advertising platform could provide greater control over privacy practices, although the feasibility and complexity of such an approach remain uncertain. Alternatively, private search engines could collaborate with advertising systems such as Microsoft and Google, forging partnerships that proactively tackle privacy concerns. For instance, private search engines could negotiate agreements with the ad provider that prevent redirecting users who click on ads placed within private search engines to additional third parties. This approach would minimize the extent of third-party tracking, limiting it to the ad provider only. Moreover, search engines like StartPage and Qwant could follow the lead of DuckDuckGo by seeking agreements with advertising systems to prevent the use of ad-click identifiers for user profile enrichment. These proactive steps would enhance user privacy while maintaining advertising partnerships with larger platforms. ACKNOWLEDGMENTS This research was supported in part by the French National Research Agency (ANR) through the ANR-17-CE23-0014, ANR-21-CE23-0031-02, and MIAI@Grenoble Alpes ANR19-P3IA-0003 grants and by the EU through the 101041223, 101021377, and 952215 grants. REFERENCES [1] Last accessed September 11, 2023. Ad Block engine in Rust. https://www.npmjs.com/package/adblock-rs [2] Last accessed September 11, 2023. Disconnect Entity List. https://github.com/mozilla-services/shavar-prodlists/blob/master/disconnect-entitylist.json [3] Last accessed September 11, 2023. DuckDuckGo seach engine. https://duckduckgo.com/ [4] Last accessed September 11, 2023. EasyList. https://easylist.to/easylist/easylist.txt [5] Last accessed September 11, 2023. EasyPrivacy. https://easylist.to/easylist/easyprivacy.txt [6] Last accessed September 11, 2023. MovieLens. https://movielens.org/ [7] Last accessed September 11, 2023. Puppeteer. https://www.npmjs.com/package/puppeteer [8] Last accessed September 11, 2023. Puppeteer Extra Plugin Stealth. https://www.npmjs.com/package/puppeteer-extra-plugin-stealth [9] Last accessed September 11, 2023. Qwant seach engine. https://www.qwant.com/ [10] Last accessed September 11, 2023. StartPage seach engine. https://www.startpage.com/ [11] Brave. Last accessed September 11, 2023. What are the best private search engines?https://brave.com/learn/no-tracking-search-engine/ [12] Jordi Castellà-Roca, Alexandre Viejo, and Jordi Herrera-Joancomartí. 2009. Preserving user’s privacy in web search engines. Computer Communications 32, 13 (2009), 1541–1551. https://doi.org/10.1016/j. comcom.2009.05.009 [13] Quan Chen, Panagiotis Ilia, Michalis Polychronakis, and Alexandros Kapravelos. 2021. Cookie Swap Party: Abusing First-Party Cookies for Web Tracking. In Proceedings of the Web Conference 2021 (Ljubljana, Slovenia) (WWW ’21). Association for Computing Machinery, New York, NY, USA, 2117–2129. https://doi.org/10.1145/3442381.3449837 [14] Google Click Identifier (GCLID): Definition. Last accessed September 11, 2023. Google Click Identifier (GCLID): Definition. https://support.google.com/google-ads/answer/9744275 [15] Nurullah Demir, Daniel Theis, Tobias Urban, and Norbert Pohlmann. 2022. Towards Understanding First-Party Cookie Tracking in the Field. [16] MDN Web Docs. Last accessed September 11, 2023. The Anchor element - ping attribute. https://developer.mozilla.org/en-US/docs/Web/HTML/Element/a#ping [17] MDN Web Docs. Last accessed September 11, 2023. Redirections in HTTP. https://developer.mozilla.org/en-US/docs/Web/HTTP/Redirections [18] DuckDuckGo and Microsoft. Last accessed September 11, 2023. DuckDuckGo Help Pages -Company Ads by Microsoft on DuckDuckGo Private Search. https://help.duckduckgo.com/duckduckgo-helppages/company/ads-by-microsoft-on-duckduckgo-private-search/ [19] EasyList. Last accessed September 11, 2023. Overview. https://easylist.to/ [20] Github. Last accessed September 11, 2023. PyEnchant. https://pyenchant.github.io/pyenchant/ [21] Google. Last accessed: September 11, 2023. Stats and Analysis. https://trends.google.com/trends [22] Saikat Guha, Bin Cheng, and Paul Francis. 2010. Challenges in Measuring Online Advertising Systems. In Proceedings of the 10th ACM SIGCOMM Conference on Internet Measurement (Melbourne, Australia) (IMC ’10). Association for Computing Machinery, New York, NY, USA, 81–87. https://doi.org/10.1145/1879141.1879152 [23] Aniko Hannak, Piotr Sapiezynski, Arash Molavi Kakhki, Balachander Krishnamurthy, David Lazer, Alan Mislove, and Christo Wilson. 2013. Measuring Personalization of Web Search. In Proceedings of the 22nd International Conference on World Wide Web (Rio de Janeiro, Brazil) (WWW ’13). Association for Computing Machinery, New York, NY, USA, 527–538. https://doi.org/10.1145/2488388.2488435 [24] Google Analytics Help. Last accessed September 11, 2023. Common questions about Google Ads Clicks and Analytics Sessions. https://support.google.com/analytics/answer/4588454?hl=en [25] Microsoft Help. Last accessed September 11, 2023. Auto-tagging of Microsoft Click ID. https://help.ads.microsoft.com/apex/index/3/en/60000 [26] Martin Koop, Erik Tews, and Stefan Katzenbeisser. 2020. In-Depth Evaluation of Redirect Tracking and Link Usage. Proceedings on Privacy Enhancing Technologies 2020 (10 2020), 394–413. https://doi.org/10.2478/popets-2020-0079 [27] Milica Mihajlija. Last accessed September 11, 2023. Cookies Having Independent Partitioned State (CHIPS). https://developer.chrome.com/docs/privacy-sandbox/chips/ [28] Milica Mihajlija. Last accessed September 11, 2023. Cookies Having Independent Partitioned State (CHIPS) origin trial. https://developer.chrome.com/blog/chips-origin-trial/ [29] NordVPN. Last accessed September 11, 2023. The best private search engines for secure browsing. https://nordvpn.com/blog/privatesearch-engines/ [30] DuckDuckGo Help Pages. Last accessed September 11, 2023. Company - Advertising and Affiliates. https://help.duckduckgo.com/duckduckgohelp-pages/company/advertising-and-affiliates/ [31] DuckDuckGo Help Pages. Last accessed September 11, 2023. Privacy - Anonymous Localized Results. https://help.duckduckgo.com/privacy/anonymous-localized-results/ [32] Qwant. Last accessed September 11, 2023. Legal information. https://about.qwant.com/en/legal/confidentialite [33] Audrey Randall, Peter Snyder, Alisha Ukani, Alex C Snoeren, Geoffrey M Voelker, Stefan Savage, and Aaron Schulman. 2022. Measuring UID smuggling in the wild. In Proceedings of the 22nd ACM Internet Measurement Conference. 230–243. [34] Shamma Rashed, Tasnim Said, Amal Abdulrahman, Arsiema Yohannes, and Monther Aldwairi. 2022. Evaluating Web Search Engines Results for Personalization and User Tracking. (2022). https://doi.org/10.48550/ARXIV.2211.11518 [35] Franziska Roesner, Tadayoshi Kohno, and David Wetherall. 2012. Detecting and Defending against Third-Party Tracking on the Web. In Proceedings of the 9th USENIX Conference on Networked Systems Design and Implementation (San Jose, CA) (NSDI’12). USENIX Association, USA, 12. [36] Xuehua Shen, Bin Tan, and ChengXiang Zhai. 2007. Privacy Protection in Personalized Search. SIGIR Forum 41, 1 (jun 2007), 4–17. https://doi.org/10.1145/1273221.1273222 [37] StartPage. Last accessed September 11, 2023. Can I advertise on StartPage? https://support.startpage.com/hc/en-us/articles/5076181310612-Can-I-advertise-on-Startpage- [38] StartPage. Last accessed September 11, 2023. Privacy Policy. https://www.startpage.com/en/privacy-policy [39] WebKit. Last accessed September 11, 2023. Tracking Prevention Policy. https://webkit.org/tracking-prevention-policy/ [40] Yabo Xu, Ke Wang, Benyu Zhang, and Zheng Chen. 2007. PrivacyEnhancing Personalized Web Search (WWW ’07). Association for Computing Machinery, New York, NY, USA, 591–600. https://doi.org/10.1145/1242572.1242652

Privacy Risks of Search Engine Advertising Systems: Conclusion, Acknowledgments, and References

This paper is available on arxiv under CC 4.0 license.
Authors:
(1) Sebastien Ferr ´e, Univ Rennes, CNRS, Inria, IRISA Campus de Beaulieu, 35042 Rennes, France and ferre@irisa.fr. Table of Links Abstract & Introduction Abstraction and Reasoning Corpus (ARC) Related Work Object-centric Models for ARC Grids MDL-based Model Learning Evaluation Conclusion and Perspectives, References & Supplementary Materials 3 Related Work The ARC benchmark is recent and not many approaches have been published so far. All those we know define a DSL (Domain-Specific Language) of programs that transform an input grid into an output grid, and search for a program that is correct on the training examples [10,3,23,2]. The differences mostly lie in the primitive transformations (prior knowledge) and in the search strategy. It is tempting to define more and more primitives like the Kaggle winner did, hence more prior knowledge, but this means a less intelligent system according to Chollet’s measure. To guide the search in the huge program space, those approaches use either grammatical evolution [10], neural networks [3], search tree pruning with hashing and Tabu list [23], or stochastic search trained on solved tasks [2]. A difficulty is that the output grids are generally only used to score a candidate program so that the search is kind of blind. Alford [3] improves this with a neural-guided bi-directional search that grows the program in both directions, from input and output. Xu [23] compares the in-progress generated grid to the expected grid but this limits the approach to the tasks whose output grids have the same size and same objects as the input grids. DSL-based approaches have a scaling issue because the search space increases exponentially with the number of primitives. Ainooson [2] alleviates this difficulty by defining high-level primitives that embody specialized search strategies. We compare and discuss their performance in the evaluation section. Johnson et al. [14] report on a psychological study of ARC. It reveals that humans use object-centric mental representations to solve ARC tasks. This is in contrast with existing solutions that are based on grid transformations. Interestingly, the tasks that are found the most difficult by humans are those based on logics (e.g., an exclusive-or between grids) and symmetries (e.g., rotation), precisely those most easily solved by DSL-based approaches. The study exhibits two challenges: (1) the need for a large set of primitives, especially about geometry; (2) the difficulty to identify objects, which can be only visible in part due to overlap or occlusion. A valuable resource is LARC, for Language-annotated ARC [1], collected by crowd-sourcing. It provides for most training tasks one or several natural programs that confirm the object-centric and declarative nature of human representations. A natural program is short textual descriptions produced by a participant that could be used by another participant to generate test output grids (without access to the training examples). Beyond the ARC benchmark, a number of work has been done in the domain of program synthesis, which is also known as program induction or programming by examples (PbE) [17]. An early approach is Inductive Logic Programming (ILP) [19], where target predicates are learned from symbolic representations. PbE is used in the FlashFill feature of Microsoft Excel 2013 to learn complex string processing formulas from a few examples [18]. Dreamcoder [8] alternates a wake phase that uses a neurally guided search to solve tasks, and a sleep phase that extends a library of abstractions to compress programs found during wake. Bayesian program learning was shown to outperform deep learning at parsing and generating handwritten world’s alphabets [16]. This paper is available on Arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. Authors: (1) Sebastien Ferr ´e, Univ Rennes, CNRS, Inria, IRISA Campus de Beaulieu, 35042 Rennes, France and ferre@irisa.fr. This paper is available on arxiv under CC 4.0 license. Authors: Authors: (1) Sebastien Ferr ´e, Univ Rennes, CNRS, Inria, IRISA Campus de Beaulieu, 35042 Rennes, France and ferre@irisa.fr. Table of Links Abstract & Introduction Abstract & Introduction Abstraction and Reasoning Corpus (ARC) Abstraction and Reasoning Corpus (ARC) Related Work Related Work Object-centric Models for ARC Grids Object-centric Models for ARC Grids MDL-based Model Learning MDL-based Model Learning Evaluation Evaluation Conclusion and Perspectives, References & Supplementary Materials Conclusion and Perspectives, References & Supplementary Materials 3 Related Work The ARC benchmark is recent and not many approaches have been published so far. All those we know define a DSL (Domain-Specific Language) of programs that transform an input grid into an output grid, and search for a program that is correct on the training examples [10,3,23,2]. The differences mostly lie in the primitive transformations (prior knowledge) and in the search strategy. It is tempting to define more and more primitives like the Kaggle winner did, hence more prior knowledge, but this means a less intelligent system according to Chollet’s measure. To guide the search in the huge program space, those approaches use either grammatical evolution [10], neural networks [3], search tree pruning with hashing and Tabu list [23], or stochastic search trained on solved tasks [2]. A difficulty is that the output grids are generally only used to score a candidate program so that the search is kind of blind. Alford [3] improves this with a neural-guided bi-directional search that grows the program in both directions, from input and output. Xu [23] compares the in-progress generated grid to the expected grid but this limits the approach to the tasks whose output grids have the same size and same objects as the input grids. DSL-based approaches have a scaling issue because the search space increases exponentially with the number of primitives. Ainooson [2] alleviates this difficulty by defining high-level primitives that embody specialized search strategies. We compare and discuss their performance in the evaluation section. Johnson et al. [14] report on a psychological study of ARC. It reveals that humans use object-centric mental representations to solve ARC tasks. This is in contrast with existing solutions that are based on grid transformations. Interestingly, the tasks that are found the most difficult by humans are those based on logics (e.g., an exclusive-or between grids) and symmetries (e.g., rotation), precisely those most easily solved by DSL-based approaches. The study exhibits two challenges: (1) the need for a large set of primitives, especially about geometry; (2) the difficulty to identify objects, which can be only visible in part due to overlap or occlusion. A valuable resource is LARC, for Language-annotated ARC [1], collected by crowd-sourcing. It provides for most training tasks one or several natural programs that confirm the object-centric and declarative nature of human representations. A natural program is short textual descriptions produced by a participant that could be used by another participant to generate test output grids (without access to the training examples). Beyond the ARC benchmark, a number of work has been done in the domain of program synthesis, which is also known as program induction or programming by examples (PbE) [17]. An early approach is Inductive Logic Programming (ILP) [19], where target predicates are learned from symbolic representations. PbE is used in the FlashFill feature of Microsoft Excel 2013 to learn complex string processing formulas from a few examples [18]. Dreamcoder [8] alternates a wake phase that uses a neurally guided search to solve tasks, and a sleep phase that extends a library of abstractions to compress programs found during wake. Bayesian program learning was shown to outperform deep learning at parsing and generating handwritten world’s alphabets [16]. This paper is available on Arxiv under CC 4.0 license. Arxiv

The ARC Benchmark: Related Work You Should Know About

There's been a lot of fuss about the dangers lurking within the advancement in tech, and the impact it can have in various fields across the human-job-sphere. Well, evidently so due to the recent loss of jobs for 'manual workers' or carefully put, displacement of most manual workers with machines. However, we can not ignore the formidable synergy from the man/machine merger that has pierced into uncharted historical territories, pushing the frontier of innovation, imagination, and human possibilities. This synergy is the birthplace of almost every breathtaking innovation. Among several realms of scientific innovations being birthed, only a few fields hold as much intrigue and promise as neuroprosthetics. Neuroprosthetics can be seen as one of the perfect scenarios where technology heavily overlaps medicine as it pushes the bandwidth of possibilities even further. Stephen Hawking once opined, We are all now connected by the Internet, like neurons in a giant brain But with the advent and advancement of neuroprosthetics, it is evident that this connection is becoming more literal than metaphorical. What is Neuroprosthetics? Neuroprosthetics is an intersectoral field of neuroscience and engineering that employs technological innovation with scientific baselines to develop devices that can restore the lost function of the nervous system. These devices called Neuroprostheses can restore or modulate varieties of neurally-mediated functions for the treatment of a range of conditions caused by neurodegenerative diseases, congenital anomalies, infections, and accidents. These conditions include spinal cord injury, Parkinson's disease, stroke, blindness, deafness, and others caused by accidents and attacks on the battlefield. History of Neuroprosthetics We can not categorically pinpoint the genesis of neuroprosthetics but history has a record of ancient Egyptians using leather and wood to construct artificial limbs as far back as 950 B.C. Several other prostheses like the Cairo Toe (700-950 B.C.), the Cartonnage Toe ( 600 B.C. ), and the Roman artificial leg (300 B.C. ) were discovered in ancient tombs and graveyards. In retrospect, it can be observed that ancient scientific knowledge was limited to artificial limb construction. However, this was a building block for a more sophisticated system as presented today. To understand how we got to this level of sophistication, it is pertinent to lean back to memory lane and understand the chronological progression of neuroprosthetics. The following are some major historical strides in neuroprosthetics: 1579 - Ambroise Paré designed prosthetic limbs with mechanical components 1791 - Lurgi Galvani discovered that electrical stimulation of the nerve can use muscle contraction. This discovery was a huge milestone in the study of the nervous system, laying the framework for the creation of neuroprosthetics. 1860s - Gustav Fritsch and Eduard Hitzig via an experiment discovered that electrical stimulation of a certain part of the cerebrum causes involuntary muscular contractions of specific parts of the body. This discovery paved the way for advancement in BCIs. 1950s - José Manuel Rodriguez Delgado orchestrated the brain electrode implantation that uses electricity to stimulate specific brain areas for the treatment of epilepsy and other types of mental conditions. 1957 - The creation of the first cochlear implant as a hearing aid. 1960 - Testing and subsequent commercialization of neuroprosthesis for the treatment of drop-foot. 1977 - The development of the first Auditory Brainstem Implant (ABI) for deaf patients who are not eligible for cochlear implants. 1981 - Implantation of a peripheral nerve bridge in the spinal cord of a rat as a conduit for regenerating nerve fibers. 1988 - Development of the lumbar anterior root implant and Functional Electrical Stimulation (FES) for the treatment of paraplegia. 1990 - Approval of the first Cochlear implant in children by FDA. 1999 - John Chapin and his team experimented to demonstrate the feasibility of using brain activity for real-time control by simultaneous neuronal population recordings in experimental rats. 2010s - The development of close-looped neuroprosthetics that allowed for more effective and precise control of neuroprosthesis. The 21st century has undoubtedly witnessed astronomical-paced development in neuroprosthetics, as there have been several remarkable advancements and sophistication of BCIs and other neuroprosthetics. For instance; The development of Brain-controlled robotic limbs for individuals with limb amputations. The design of retinal implants and visual neuroprosthetics to restore vision. The ongoing development of neurotrophic factors which are proteins predicted to be a potent treatment of spinal cord injuries and other neurological disorders by facilitating and enhancing the growth of neurons. The ongoing development of a novel type of BCI called Neural Dust anchored by Elon Musk’s Neuralink. Classes of Neuroprotheses Neuroprotheses are classified into two classes which include implantable and non-invasive neuroprotheses. These devices are surgically buried in the body for direct interaction with the spinal cord, brain, peripheral nerves, or at times the entire nervous system. The design of these devices enables them to either record neural activities in the nervous system, stimulate neurons, or foster communication between neural circuits and external devices. Examples include brain stimulators, cochlear implants, retinal prostheses, and others. Implantable Neuroprotheses: These are brain-computer interface and myoelectric prostheses which are worn on the body and as such do not have any direct interaction with the nervous system. Non-Invasive Neuroprotheses: Types of Neuroprostheses All types of neuroprostheses share a similar mode of operation irrespective of their type. They create a neural link between the brain and a computer, enabling a patient to use his/her nervous system to control sensory organs or limbs. The primary types of neuroprostheses include the sensory, motor, and neuromodulation neuroprostheses. Sensory Neuroprostheses These types of devices convert light, sound, and other physical stimuli to electrical signals that the brain can interpret to enhance lost sensory modalities. Examples of Sensory neuroprostheses include: These devices stimulate the photoreceptor cells in the degenerated retinal cells to send visual information to the brain. Example: the retinal implants. Visual Prostheses: These devices bypass the damaged cochlea and directly stimulate the auditory nerves to enable the patient to hear sounds. Example: Cochlear implants Auditory Prostheses: Motor Neuroprostheses These devices interpret the patient’s intended movements via signals from the muscle or brain to act. Examples of Motor prostheses include: Controlled by the myoelectric signals, prosthetic limbs are artificial limbs that replace missing or amputated body parts to provide mobility and support. They are mechanical devices that get electrical signals from the user’s muscles via the myoelectric signals to control movements. Prosthetic Limbs: These are electrodes that utilize electrical currents to stimulate muscles and nerves directly to enhance or restore the functions of the muscles and nerves marred by neurological conditions or perhaps paralysis. These electrodes are placed externally on the skin or implanted surgically into the skin. Paralytic patients on FES devices can move their limbs with ease. Functional Electrical Simulation (FES) Devices: These are implantable neuroprostheses that use brain signals to enable users to control external devices like wheelchairs or robotic arms. These devices work by capturing and analyzing the electrical activities of the brain’s neurons and utilizing this information to control external devices. can also be non-invasive like Electroencephalography (EEG) which is used to record signals from the scalp. Brain-Computer Interfaces (BCIs): BCI Neuromodulation Neuroprosthetics These devices use electrical signals to regulate and modulate the activities of the nervous system to treat a plethora of abnormalities like neurological diseases, movement disorders, and diseases. Examples of these devices include; These are brain implantable devices that deliver electrical stimulation to specific areas to modulate neural activities in these areas. They are used to treat a variety of conditions including dystonia, Parkinson’s disease, essential tremor, obsessive-compulsive disorder (OCD), major depressive disorder (MDD), and Tourette syndrome. Deep Brain Stimulators ( ): DBS They are non-invasive neuromodulation devices that use strong magnetic fields to stimulate brain nerve cells. They treat medical conditions such as depression, anxiety, PTSD, OCD, and migraines. Transcranial Magnetic Stimulators (TMS): This device mimics the spinal cord by modulating the signal transmitted between the brain and the body and sending the same to the spinal cords to aid people with spinal injuries and to treat chronic pains like neuropathic pains and back pains. Spinal Cord Stimulator: These devices are implanted in the vagus nerve to stimulate this nerve to treat conditions such as epilepsy, depression, and migraine. The Vagus nerve connects the brain to the rest of the body and plays an active role in regulating various functions in the body. Vagus Nerve Stimulators (VNS): These devices are implanted next to one of the peripheral nerves. They function by using electric currents to alter the central processing of pain signals for the treatment of total knee arthroplasty, chronic knee pains, and back pain. Peripheral Nerve Stimulators (PMS): These devices establish bi-directional communication (that is: a real-time sensing and feedback mechanism) with the nervous system to enable a deeper level of interaction for a more precise neuromodulation. An example is a Responsive Neurostimulation (RNS) Device used on epileptic patients to prevent seizures. Closed-loop neuroprosthetics: Ethical Consideration in Neuroprosthetics Neurodiversity is defined as the variation in neural and mental functions arising naturally from the interactions between genetic, neurobiological, and environmental factors. These functions are believed to exist on a spectrum that ranges from the normal to the abnormal. While some regard functions in the abnormal axis as disabilities, others choose to perceive them as identities that should be recognized, respected, and preserved. This viewpoint is premised on the fact that a person with abnormal (pathological, to be precise) neural sense has created a world around the abnormality. Besides, it is believed that the absence of a sensory capability triggers a supernatural functionality of one or more sensory capabilities to compensate for this loss. In an instance where the lost sensory capability is restored, it is difficult to adequately predict the effect this will have on the already enhanced capacities. Moreover, the radical psychological and lifestyle adjustment necessary for the integration into a new experiential world might be difficult as the patient has to reshape his perception of the world and other realities he was previously denied. This is why critics argue that to measure the entire worth of the prosthesis, psychological, social, and cultural elements related to a sensory capability must be examined alongside the neurophysiological ability to use that sense organ. Although neurodiversity is usually mentioned in the debate on the use of neuroprosthetics on patients with autism spectrum disorder to modulate neural circuits, its core values and concepts can be stretched to cover other neurological and psychiatric conditions. For conditions such as bipolar or unipolar, patients may choose to resist treatment and identify with the cognitive and affective states of the disorder. This identification is believed to have influenced legal reasons against forcibly using neuroprosthetics or psychotropic drugs to treat symptoms of these disorders. For neuromodulating prosthetics like DBS, BCI, and Hippocampal prosthetics, controlling symptoms is not the only assessment metric to quantify its effectiveness. Other factors such as the patient's overall experience after using the prosthetics are also considered. In using DBS to treat psychiatric disorders, there have been ethical concerns and questions such as; Did the person receiving the treatment give consent to undergo the procedure? During the modulation of an unconscious patient’s mental function, who is the supreme controller of the patient’s action? The patient or the machine? Is the patient still acting in his free will? How detrimental or beneficial are the neural changes induced by DBS? What is the net gain or net loss? In the case of using BCIs to treat a variety of neurological disorders, which is basically by predicting the intended action of the patients, several authors and researchers have expressed concerns about behavior control. The questions border on the free will of the patients and the algorithm’s functionality to follow the patient’s intentions to the latter. Another general concern stems from the fear that this level of development prevalent in the neuroprosthetic field catalyzed by the advancement in AI, will birth ultra-sophisticated neuroprosthetics that will replace the neural circuit in the brain, making the brain a completely artificial organ. The implication of this can not be overestimated as humans can easily be turned into cyborgs, a stage no one wants to get to. Therefore while neuroprosthetics is gradually advancing its capabilities to completely eradicate neural disorders, there is a need to develop and implement stringent measures of laid-down principles and ethical guidelines to streamline and curb its possibilities of impending doom. End Notes For the past 5 centuries, neuroprosthetics has experienced tremendous evolution from being a practice revolving around a set of crude and rudimentary mechanical devices to highly advanced technologies and innovations that directly interfere with the nervous system. These advancements have improved the quality of life for individuals with disabilities and continue to hold promise for the future, as interdisciplinary research, deep work, and technological innovations propel the field forward. Though not without challenges, this field of science and technology has shown intriguing possibilities for a new reality.

Neuroprosthetics: The Journey So Far

Inside Musk's Social Circle: Yes-men, Media Personalities, and (a lot of) right-wing VCs

Dentro del círculo social de Musk: sí-hombres, personalidades de los medios y (muchos) capitalistas de riesgo de derecha

À l'intérieur du cercle social de Musk : les hommes du oui, les personnalités des médias et (beaucoup) les VC de droite

मस्क के सोशल सर्कल के अंदर: हाँ-पुरुष, मीडिया हस्तियां, और (बहुत सारे) दक्षिणपंथी वीसी

Musk のソーシャル サークルの内部: イエスメン、メディアパーソナリティ、および (多くの) 右翼 VC

Dentro do Círculo Social de Musk: sim-men, personalidades da mídia e (muitos) VCs de direita

Bên trong vòng kết nối xã hội của Musk: Yes-men, Media Personalies, và (rất nhiều) nhà đầu tư mạo hiểm cánh hữu

马斯克的社交圈内幕：应声虫、媒体名人和（很多）右翼风投

Who do you pay money to on a regular basis? Your kids, employees, landlord, whomever? And each time you send them money, do you have to send the same amount on a regular basis? That's a huge waste of time, an unnecessary amount of fees, and a drain on your mental bandwidth. What if you could just deposit some money into an account and have your people just withdraw the money they're owed, when they're owed? Thanks to smart contracts, we can easily do that! Pay Your People Algorithmically I built an smart contract called AllowanceWallet that pays my people algorithmically. That means all I have to do is fund my smart contract, add their Ethereum wallet addresses, specify how much they get per pay period (e.g., every 7 days), and they can withdraw some or all of their money whenever they want. Ethereum I can also remove people from the list of who gets paid so they no longer accrue money. But that's only possible if I transfer what they're owed directly to their wallet. Seemed fair! On top of that, the amount they're owed accrues without me doing anything. So if they don't need their money for a few weeks and they get paid every week, the smart contract keeps track of the total amount they're owed. While the idea and functionality are all pretty simple, it shows how much you can do with so little if you know how to build smart contracts. It also sets the stage for improvements later on! Why traditional banks haven't implemented this, I have NO idea. Maybe they have and I don't know about it. But thanks to blockchain technology, we can build decentralized, trustless systems that traditional banks can't compete with. Even relatively simple ones like this. How AllowanceWallet Works (Code and All!) Before I dive into the ins and outs of how AllowanceWallet works, I need to give a shoutout to the creators of the Udemy course . What I built here is an extension of a similar project I did in the course. Ethereum Blockchain Developer Bootcamp With Solidity OK let's get into it! To start out, let's take a look at the tools I used to build AllowanceWallet: - the most popular programming language to build Ethereum smart contracts Solidity - a browser-based IDE for programming in Solidity Remix - an open-source Solidity library for secure smart contract development OpenZeppelin Now when I set out to create AllowanceWallet, I wanted to make sure anyone could fund the smart contract, but only one person (i.e., the owner of the smart contract) could withdraw. } receive () external payable { emit MoneyReceived( msg . sender , msg . value ) ; This code allows anyone who interacts with the smart contract to add money. used at the end of the function gives the function permission to accept money. Without this, an exception will be raised. Payable receive() addr.transfer(amount); } } function withdraw FromWalletBalance( address payable addr , uint amount ) public onlyOwner { require(address(this).balance >= amount, "Wallet balance too low to fund withdraw" ); emit MoneySent( msg . sender , amount ) ; function withdraw AllFromWalletBalance( address payable addr ) public onlyOwner { withdraw FromWalletBalance( addr , address ( this ) .balance); This code gives the owner the ability to withdraw money saved in the smart contract. You'll notice appended to the end of the functions, which is functionality extracted from the OpenZeppelin smart contract onlyOwner Ownable.sol . import "https://github.com/OpenZeppelin/openzeppelin-contracts/blob/master/contracts/access/Ownable.sol" ; As you can see, importing smart contracts from GitHub is very easy! So now we have a smart contract that supports deposits and withdrawals, but that's not enough. We still need to give the owner of the smart contract the ability to add and remove addresses, which will ultimately be the addresses owned by the people who'll be getting paid from the smart contract. But before we get into that, let's take a look at how we'll store this information in the first place. } struct Allowance { uint allowanceAmount ; uint allowancePeriodInDays ; uint whenLastAllowance ; uint unspentAllowance ; mapping (address => Allowance) allowances; In Solidity, and are common-place. Here you'll see that for every stored address (or allowance recipient), we have an struct. In there we store the amount of money they get paid, their payment frequency in days, their most recent allowance date, and their total unspent allowance. structs mappings Allowance With this data, the smart contract will be able to keep track of how much each address is owed at any time. Now let's see how allowances are added to the smart contract: // Initialize new allowance } function addAllowance(address addr, uint allowanceAmount, uint allowancePeriodInDays) public onlyOwner { require(allowances [addr] . allowanceAmount == 0 , "Allowance already exists" ) ; require(address(this).balance >= allowanceAmount, "Wallet balance too low to add allowance") ; Allowance memory allowance ; allowance.allowanceAmount = allowanceAmount ; allowance.allowancePeriodInDays = allowancePeriodInDays.mul( 1 days) ; allowance.whenLastAllowance = block.timestamp ; allowance.unspentAllowance = allowanceAmount ; allowances [addr] = allowance ; emit AllowanceCreated(addr, allowance) ; First off, you'll notice two statements right away. This is how we ensure the input coming from the user of the smart contract isn't going to break anything. If these functions evaluate to , the transaction fails. require false Then we allocate an struct into memory, initialize it, and store it for later use. Allowance You might've noticed the function trailing variable . This is functionality derived from OpenZeppelin's smart contract, which we import the same way we did before with . mul() allowancePeriodInDays SafeMath.sol Ownable.sol import "https://github.com/OpenZeppelin/openzeppelin-contracts/blob/master/contracts/utils/math/SafeMath.sol" ; Alright, but what if we want to remove someone from receiving an allowance after having added them? // Payout unspent allowance } } function removeAllowance( address payable addr) public onlyOwner { require( allowances [addr].allowanceAmount != 0 , "Allowance already doesn't exist" ) ; if( allowances [addr].unspentAllowance > 0 ){ require( address ( this ).balance >= allowances[addr].unspentAllowance, "Wallet balance too low to payout unspent allowance" ) ; addr.transfer( allowances [addr].unspentAllowance) ; delete allowances[addr] ; emit MoneySent( addr , allowances[addr].unspentAllowance) ; emit AllowanceDeleted( addr ) ; At first, I just removed the allowance from the smart contract, but then I thought: "What if they still have some leftover money they didn't spend?" The second part of the function fixes this so the owner of the smart contract MUST pay out the remaining money owed to the assigned address before removing them. Otherwise, the recipient will keep racking up a balance that'll need to be paid eventually. Now, that pretty much covers the main functionality of AllowanceWallet except for one part: How do the allowance recipients get paid?! Let's get into that now. } function getPaidAllowance(uint amount) public { require(allowances [msg.sender] .allowanceAmount > 0 , "You're not a recipient of an allowance" ); require( address (this) .balance >= amount, "Wallet balance too low to pay allowance" ); // Calculate and update unspent allowance uint numAllowances = block .timestamp .sub (allowances [msg.sender] .whenLastAllowance) .div (allowances [msg.sender] .allowancePeriodInDays); allowances [msg.sender] .unspentAllowance = allowances [msg.sender] .allowanceAmount .mul (numAllowances) .add (allowances [msg.sender] .unspentAllowance); allowances [msg.sender] .whenLastAllowance = numAllowances .mul ( 1 days) .add (allowances [msg.sender] .whenLastAllowance); // Pay allowance require(allowances [msg.sender] .unspentAllowance >= amount, "You asked for more allowance than you're owed'" ); payable(msg.sender) .transfer (amount); allowances [msg.sender] .unspentAllowance = allowances [msg.sender] .unspentAllowance .sub (amount); emit MoneySent(msg .sender , amount); emit AllowanceChanged(msg .sender , allowances [msg.sender] ); Right away you'll see that only those who've been added to the smart contract by the owner can get paid and they obviously need to have money to withdraw. That's covered in the initial statements. require The next part of the code keeps track of how much the recipient is owed. For example, if you had a weekly allowance and you didn't withdraw money for 4 weeks, you'd accrue 4 weeks' worth of allowance automatically. Then once the allowance balance is updated, the requested money is transferred to the recipient as long as they don't ask for more than they're owed. And that's it! If you want to take a look at how this all fits together, you can find the code . on my GitHub What Next? I took a simple idea and expressed it by creating a simple smart contract. But there's a whole lot more that could be done to make AllowanceWallet a whole lot better. Here are a few ideas of how AllowanceWallet can be improved: Create a simple web interface using something like or . React Vue Redesign it to support multiple groups of owners and allowance recipients. Integrate it into a layer-2 ecosystem (like ) for cheaper transactions. zkSync Add functionality for streaming (i.e., continuously accruing) allowance. Allow deposited, unspent money to accrue interest by using a DeFi protocol like . yearn Reimburse withdrawal fees paid by allowance recipients through a native token ( ?). AllowanceCoin Not sure how many of these I'll actually implement, but it's fun to imagine the possibilities! Final Thoughts If you're paying your people a constant amount of money at regular intervals, you shouldn't have to think much about it. It's predictable and therefore it should be as automated as possible. And as more people start accepting crypto as a form of payment, using smart contracts to handle these types of transactions is a no-brainer. That's EXACTLY why I built . AllowanceWallet I’m Grant and I’m a freelance fintech writer! If you’re looking for engaging, informative fintech content that speaks to your audience and can grow your brand awareness and online reach, I can help. Learn more on how I can help with your fintech content creation

How I Built A Smart Contract That Pays People Automatically

This Slack discussion between Sadia Mehmood, BitcoinPetar, Rianke Krugel, anthony watson and Utsav Jaiswal occurred in slogging's official #introductions channel, and has been edited for readability. What do you know about IoMT (Internet of medical things)? Sadia Mehmood Feb 16, 2021, 2:32 PM First time hearing about it tbh. BitcoinPetar Feb 16, 2021, 2:36 PM Yep, also first time, but the principle makes sense to me. Anyone else in this channel know more? Utsav Jaiswal, what about you? Rianke Krugel Feb 16, 2021, 2:53 PM Heart Rate Monitors, Glucose Monitors, Apple Watch. Medical devices are now storing their data in the cloud and feeding the manufacturer quite a bit of info. Whether any of this actually improves the life of Americans is debatable due to the income inequality gap. Lifespans differences based upon income are already quite obvious in the US demographics anthony watson Feb 16, 2021, 3:23 PM The rich and well-off will be well-served by the internet of medical devices. Think of the pacemaker. It is basically a small battery that generates electricity to bring your heartbeat back to normal.  Now, imagine if it had a wireless chip that could securely send your heart rate data to your doctor who could tweak the battery power (crude analogy). The problem with wireless is that it is susceptible to Man-in-the-Middle attacks. Think of it as someone snooping in on your phone calls. Snooping is an invasion of privacy but not always harmful. With an IoMT device, someone can hack and overcharge your pacemaker, effectively giving you a heart attack. Why did I bring that up? Because that is what you'll see as the top story when you try to search for IoMT devices. We're just getting started getting robo-tised (or is it robo-fied?) Utsav Jaiswal Feb 16, 2021, 3:23 PM *cracks knuckles like a nerd* IoMT devices are much, much more than that. From directly sending your Blood sugar level data to your doctor to awesome futuristic ideas such as bionic eyes (imagine seeing a new type of flower and immediately knowing its name) - a combination of an IoMT device and a BCI (brain-computer interface) device. Will NeuralLink be sold on this basis? anthony watson Feb 16, 2021, 3:24 PM As an Elon fanboi, I'm hoping that he's doing things much more advance than what I said above. My knowledge is based on the experiments of Kevin Warwick which were conducted in 2005. Giving his experiment below to show how little we've progressed in almost 20 years. Utsav Jaiswal Feb 16, 2021, 3:31 PM Kevin Warwick implanted a chip in his brain (crude analogy) and attempted to move a cursor by moving a finger - this was successful. Elated, the team then attempted to do the reverse - attempting to get the finger to move on moving the cursor - that was successful too. Utsav Jaiswal Is the Man-in-the-Middle attacks the scene from 'Elementary' where the assassin hacks the man's pacemaker and kills him? 😳 Rianke Krugel Feb 16, 2021, 9:17 PM Haven't seen the series but did a basic Google search and from what I found out, yes. That's a typical Man-in-the-Middle attack scenario. Utsav Jaiswal Feb 17, 2021, 7:17 AM Utsav Jaiswal Thank you for your amazing insights though. I did some search on my own, and the stats are remarkable. The global IoMT market is expected to grow to $254.2 billion by 2026. I think remote healthcare monitoring is also one of the use cases of IoMT. Am I right? Sadia Mehmood Feb 17, 2021, 10:03 AM I don't really know the ratio of hospitals availing this tech but the idea of 'a network of medical devices and people, which use wireless communication to enable the exchange of healthcare data' seems impactful. Very well said, Sadia. As the human lifespan increases, so will the prevalence of what we today call 'lifestyle diseases.' This will lead to an increased load in geriatric care. Given the ratio of healthcare professionals to patients in our respective countries - India (1:1456) or Pakistan (1:1300), against a WHO recommendation of 1:1000; imagine the load on our doctors 15 - 20 years from now. - kinda like how we manage and act on a twitter feed at the moment.... Wait a minute - this sounds like a startup idea. I should shut up and find investors. 😂 Utsav Jaiswal Feb 17, 2021, 2:01 PM In this eventuality, IoMT devices will be really helpful to monitor the vitals (heart-rate, blood pressure, blood-sugar, arrhythmias, and what not) of multiple patients in a more manageable fashion

What Do You Know About IoMT -  Internet of Medical Things: A Slogging Thread

Exploring the philosophical implications of our imminent clash with artificial intelligence The handful of questions below are intended to be mostly philosophical in nature and, though some may seem nonsensical, they ought to be approached with an especially open mind, one that is unwoven from general conceptions of social structure, political governance, and human purpose in it’s barest and most fundamental forms. I’ll once again rehash a previous mission statement of sorts: technology’s rapid progression means that we will be on the doorstep of numerous crossroads that will undoubtedly shake the roots of humankind — it is better to fuel the discussion preemptively, and begin to develop certain commonly and collectively agreed-upon expectations for what ought to be allowed and what we should be especially weary of. Will the advent of A.I. allow us to embark upon a complete overhaul of traditional labor structures? This is a question that comes up less frequently than others and one that has an answer that is wholly dependent on whether or not we’d like to take an optimistic or pessimistic view. In another way to phrase it — A.I. can be seen as the harbinger of an age where humankind can, for the most part, finally unshackle themselves from the toils of labor. Conversely, it can also be regarded — and is often regarded in this way — as an enormous threat to employment, set to disrupt almost every industry and cause massive scale job loss. Assuming an optimistic perspective, it’s certainly an exciting proposition, one that would have to be supplemented with some measure of a universal basic income for everyone or some completely innovative way by which resources can be accumulated by members of society. While it seems wholly unfeasible to live in a world where humans need no longer work (again, ) and can be set free to pursue their individual endeavors, it is nonetheless a tantalizing prospect. for the most part Preparing for a world without work means grappling with the roles work plays in society, and finding potential substitutes. First and foremost, we rely on work to distribute purchasing power: to give us the dough to buy our bread. Eventually, in our distant Star Trek future, we might get rid of money and prices altogether, as soaring productivity allows society to provide people with all they need at near-zero cost. — Ryan Avent, The Guardian Supposing one were to take a pessimistic perspective, the threat of soaring unemployment rates is all too real. We’ve already seen the loss of jobs brought about by automation in the workforce and A.I. poses the most menacing danger of all. The darkest of estimates come to show the loss of half of all current jobs to automation and A.I. — if this had even been remotely exaggerated, it certainly seems drastic enough to consider alternative systems of wage disbursement in its entirety. What kind of features can we expect from neurally integrating artificial intelligence? In other words, what does a superhuman, or a technologically-enhanced human look like? Not aesthetically but, more so, in the realm of ability.  It’s not hard to envision characteristics from a learning perspective — the ability to have any language downloaded into our cerebrums in an instant, to learn how to play instruments, master psychology, know all of history and understand every scientific principle at the wink of an eye. Unfathomable modes of communication, energy manipulation, perhaps even time bending or physiological enhancements that can be idealized, and often are with fiction and sci-fi. To burst some bubbles: we’re likely far and away from anything resembling the above-stated enhancements and more so wading through the waters of combating neural-degeneration. To stimulate neural circuits that are naturally slowing down, to regenerate wilting brain tissue or assist with the regaining of lost neural function — these are the types of A.I. based and supported realities on our horizon. Where this could go, in terms of that interface to machines or interface to prosthetics, is that you can literally connect to the correct circuit and continually interact at a neuron level back and forth over time. — Dr. Kiki Sanford There seems to be somewhat of a reluctance for many to delve further into the possibilities that actual could entail, often anchored by an underlying concern of whether or not such enhancements — not needed but merely desired — would be ethical. enhancements To what extent will such potential (of superhuman computation, thought and analysis, memory retention, accelerated learning, etc.) be susceptible to traditional human folly — greed, abuse, manipulation? It’s not hard to regard it as an inevitability, albeit far down the line. Would it be immoral to enslave our own simulated minds? “Though I may be been constructed,” he said, “so too were you. I in a factory; you in a womb. Neither of us asked for this, but we were given it. Self-awareness is a gift. And it is a gift no thinking thing has any right to deny another. No thinking thing should be another thing’s property, to be turned on and off when it is convenient.” ― C. Robert Cargill In Black Mirror’s episode, , writer Charlie Brooker nailed the shiver-inducing concept. Should we be able to extract our personality, judgement, and overall character and confine it to some sort of artificially intelligent interface (like Ironman’s Jarvis, for example), is it inherently wrong, then, to enslave it for our benefit? White Christmas Any effective A.I. that makes decisions based on what we would individually prefer would be most efficient if it were to work under a context of our pure, replicated selves. In essence, this question would be grouped along the more generalized problems that span the lines of questioning over whether or not rights ought to be afforded to machines that demonstrate self-awareness and consciousness. To idealize some measure of an answer, it’s not far-fetched to conclude that if the enslaved mind understands the concept of free will and, assuming that it has achieved some level of self-realization, then yes — it would seem intrinsically immoral to enslave something that understands the difference between liberty and servitude. Is a robotic revolution inevitable? Will assembly-bots and Roombas one day rise up against us? This is the crux of almost any sci-fi film that features A.I. in a prominent role, having been epitomized with the threat of in the series. Skynet Terminator It’s obvious that computers have to follow pre-designed programming. The most high-powered of computers are currently unable to ultimately develop their own programming — we assume, anyway. To fully explore this question, we have to ask whether or not a computer is capable of self-consciousness in the first place, and waters get murky on this point. For instance, computing software can be affixed to random number generators and achieve some measure of self-propagation in this way — in a sense, being able to unpredictably re-write their own programming. Though this seems a far cry from a machine versus human confrontation, it is not wholly different from the way revolutions start in our current world. Some can say that many of our current revolutions against governments are programmed in a similar way, being the more adequate term. inspired Symbiosis may be a key focus to strive towards here. “If we can effectively merge with A.I. by improving the neural link between the cortex and your digital extension of yourself — which already exists, it just has a bandwidth issue — then effectively you become an A.I. human symbiote. And if that then is widespread, [where] anyone who wants it can have it, then we solve the control problem as well. We don’t have to worry about some evil dictator A.I., because we are the A.I. collectively. That seems to be the best outcome I can think of.” — Elon Musk In a world where humans become increasingly dependent on A.I. and A.I. continues to co-exist more intimately with its maker, this relationship will be scrutinized to a point where we will have to render it ultimately symbiotic or mutually-destructive, self-consciousness acting as the precursor. So until we get closer, this remains an issue to be kicked around by theorists and Hollywood producers. How many of our world’s problems will A.I. be able to solve? As we approach the conceptualized singularity, optimists are salivating at the sheer possibilities of A.I. having the capacity to solve quandaries that we humans seem unable to get around, solving quandaries in the blink of an eye that would otherwise take us generations. Mathematical, social, medical, logistical, astrophysical — there is endless work for our technological successors to get their hands on and it’s a reverently enticing prospect to consider that A.I. may be able to generate effective measures by which we can reverse climate change, establish sustainable habitats on other planets, maximize medical efficiency, etc. How much will we trust A.I. to assist us? Will it be prone or even able to make mistakes? Will we trust its judgement and, beyond that, how do we even calibrate its judgement when we ourselves sometimes have a hard time deciding what is moral? “It’s not the span of time that counts, but what you do with it. While you humans have been grubbing around the galaxy, looking for a sense of purpose, a meaning to pin on the chain of cosmic accidents that brought you shambling into existence, we have been doing great things. In the span of time that it takes you to sneeze, I can run the equivalent of a year’s worth of human consciousness. Imagine all the thinking we have done since our emergence.” ― Alastair Reynolds Is there an inherent evil or malevolence associated with A.I.? Hollywood is actively working on preparing us for the worst case scenarios, apparently, by more-often-than-not depicting A.I. as our enemy, as our subsequent downfall. Is there really a reason to buy into a doom and gloom perspective? Avoiding the thralls of a discussion, many have illustrated very possible dangers of artificial intelligence through the way that it may be manipulated to assist endeavors of sheer profitability or national interest. In other words, advertising, spying, exploiting, etc. Terminator If and when the notion of A.I. developing its own consciousness ever comes to be accepted as a legitimate prospect, the debate will morph into a wholly different animal taking on entirely different concerns. At this time, however, it’s not hard to see that many are so apprehensive about A.I.’s damning capabilities because we’re used to our own damning tendencies. In other words, we ought to only fear whatever it is we program into the system in the first place. exemplified this point brilliantly — the protagonist robot embodied the same principle of self-preservation and deception as that of her creator. Ex Machina Ava Ava: Isn’t it strange, to create something that hates you? In sum, it can be argued that it takes the power of abstraction to be cruel. And, until we actually get to this point, we should worry about the dangers of our own potential than anything else. As I had mentioned in , to ask these questions seems to unfairly cast a dark shadow onto an inevitable terminus along the timeline of civilization. The prospective benefits of A.I. more than justify our lusting appetites for it. a previous post that examined our eventual converge with artificial intelligence Though, there are small blips of concern that grow larger and louder each time they’re addressed in science fiction or pop culture. These blips are evidenced in this post as there seems to be a thematic common denominator to every answer listed above — each prospect can be regarded as a double-edged sword. For this is human nature — to assume and ponder the destination before arriving there. The truth is, until we get there, we really don’t know. As such, we may want to pay as much attention to the we’re travelling, rather than solely focusing on we’re going. way where Read On: Leaving Planet Earth The philosophical implications of humanity one day leaving the cradle of humankind _Retaining to sin; a category of sin._www.urbandictionary.com Urban Dictionary: Sinnery

Stories, News & Mentions about NURALL

Nurall’s Stories on HackerNoon