At face value, this essay is about understanding a fairly esoteric governance tool called compute thresholds. However, in order to grapple with whether these thresholds will achieve anything, we must first understand how they came to be. This requires engaging with a decades-old debate at the heart of computer science progress, namely, is bigger always better? Hence, this essay may be of interest not only to policymakers and the wider public but also to computer scientists interested in understanding the role of compute in unlocking breakthroughs. Does a certain inflection point of compute result in changes to the risk profile of a model? This discussion is increasingly urgent given the wide adoption of governance approaches that suggest greater compute equates with higher propensity for harm. Several leading frontier AI companies have released responsible scaling policies. Both the White House Executive Orders on AI Safety (EO) and the EU AI Act encode the use of FLOP or floating-point operations as a way to identify more powerful systems. What is striking about the choice of compute thresholds to-date is that no models currently deployed in the wild fulfill the current criteria set by the EO. This implies that the emphasis is often not on auditing the risks and harms incurred by currently deployed models - but rather is based upon the belief that future levels of compute will introduce unforeseen new risks. A key conclusion of this essay is that compute thresholds as currently implemented are shortsighted and likely to fail to mitigate risk. Governance that is overly reliant on compute fails to understand that the relationship between compute and risk is highly uncertain and rapidly changing. It also overestimates our ability to predict what abilities emerge at different scales. This essay ends with recommendations for a better way forward.
从表面上看，这篇文章的主题是了解一种名为计算阈值的相当深奥的治理工具。然而，为了探讨计算阈值是否能实现任何目标，我们必须首先了解计算阈值是如何产生的。这就需要讨论计算机科学进步的核心问题，即 "是否越大就越好？"这一长达数十年之久的争论。因此，这篇文章不仅对政策制定者和广大公众有意义，而且对有兴趣了解计算在实现突破中的作用的计算机科学家也有意义。计算的某个拐点是否会导致模型的风险状况发生变化？这一问题的讨论越来越紧迫，因为人们普遍采用的治理方法认为，计算能力越强，造成危害的可能性就越大。一些领先的前沿人工智能公司已经发布了负责任的扩展政策。白宫关于人工智能安全的行政命令（EO）和欧盟人工智能法案都将 FLOP 或浮点运算作为识别更强大系统的一种方式。迄今为止，在计算阈值的选择上令人震惊的是，目前部署在野外的模型都不符合 EO 目前设定的标准。这意味着，重点往往不在于审核当前部署的模型所带来的风险和危害--而是基于这样一种信念，即未来的计算水平将带来不可预见的新风险。本文的一个重要结论是，目前实施的计算阈值是短视的，很可能无法降低风险。过度依赖计算的治理方式未能理解计算与风险之间的关系是高度不确定和快速变化的。同时，它也高估了我们预测不同规模下可能出现的风险的能力。本文最后提出了更好的发展建议。
{"title":"On the Limitations of Compute Thresholds as a Governance Strategy","authors":"Sara Hooker","doi":"arxiv-2407.05694","DOIUrl":"https://doi.org/arxiv-2407.05694","url":null,"abstract":"At face value, this essay is about understanding a fairly esoteric governance\u0000tool called compute thresholds. However, in order to grapple with whether these\u0000thresholds will achieve anything, we must first understand how they came to be.\u0000This requires engaging with a decades-old debate at the heart of computer\u0000science progress, namely, is bigger always better? Hence, this essay may be of\u0000interest not only to policymakers and the wider public but also to computer\u0000scientists interested in understanding the role of compute in unlocking\u0000breakthroughs. Does a certain inflection point of compute result in changes to\u0000the risk profile of a model? This discussion is increasingly urgent given the\u0000wide adoption of governance approaches that suggest greater compute equates\u0000with higher propensity for harm. Several leading frontier AI companies have\u0000released responsible scaling policies. Both the White House Executive Orders on\u0000AI Safety (EO) and the EU AI Act encode the use of FLOP or floating-point\u0000operations as a way to identify more powerful systems. What is striking about\u0000the choice of compute thresholds to-date is that no models currently deployed\u0000in the wild fulfill the current criteria set by the EO. This implies that the\u0000emphasis is often not on auditing the risks and harms incurred by currently\u0000deployed models - but rather is based upon the belief that future levels of\u0000compute will introduce unforeseen new risks. A key conclusion of this essay is\u0000that compute thresholds as currently implemented are shortsighted and likely to\u0000fail to mitigate risk. Governance that is overly reliant on compute fails to\u0000understand that the relationship between compute and risk is highly uncertain\u0000and rapidly changing. It also overestimates our ability to predict what\u0000abilities emerge at different scales. This essay ends with recommendations for\u0000a better way forward.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"2016 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141571837","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Constrained optimization underlies crucial societal problems (for instance, stock trading and bandwidth allocation), but is often computationally hard (complexity grows exponentially with problem size). The big-data era urgently demands low-latency and low-energy optimization at the edge, which cannot be handled by digital processors due to their non-parallel von Neumann architecture. Recent efforts using massively parallel hardware (such as memristor crossbars and quantum processors) employing annealing algorithms, while promising, have handled relatively easy and stable problems with sparse or binary representations (such as the max-cut or traveling salesman problems).However, most real-world applications embody three features, which are encoded in the knapsack problem, and cannot be handled by annealing algorithms - dense and non-binary representations, with destabilizing self-feedback. Here we demonstrate a post-digital-hardware-friendly randomized competitive Ising-inspired (RaCI) algorithm performing knapsack optimization, experimentally implemented on a foundry-manufactured CMOS-integrated probabilistic analog memristor crossbar. Our solution outperforms digital and quantum approaches by over 4 orders of magnitude in energy efficiency.
{"title":"Energy Efficient Knapsack Optimization Using Probabilistic Memristor Crossbars","authors":"Jinzhan Li, Suhas Kumar, Su-in Yi","doi":"arxiv-2407.04332","DOIUrl":"https://doi.org/arxiv-2407.04332","url":null,"abstract":"Constrained optimization underlies crucial societal problems (for instance,\u0000stock trading and bandwidth allocation), but is often computationally hard\u0000(complexity grows exponentially with problem size). The big-data era urgently\u0000demands low-latency and low-energy optimization at the edge, which cannot be\u0000handled by digital processors due to their non-parallel von Neumann\u0000architecture. Recent efforts using massively parallel hardware (such as\u0000memristor crossbars and quantum processors) employing annealing algorithms,\u0000while promising, have handled relatively easy and stable problems with sparse\u0000or binary representations (such as the max-cut or traveling salesman\u0000problems).However, most real-world applications embody three features, which\u0000are encoded in the knapsack problem, and cannot be handled by annealing\u0000algorithms - dense and non-binary representations, with destabilizing\u0000self-feedback. Here we demonstrate a post-digital-hardware-friendly randomized\u0000competitive Ising-inspired (RaCI) algorithm performing knapsack optimization,\u0000experimentally implemented on a foundry-manufactured CMOS-integrated\u0000probabilistic analog memristor crossbar. Our solution outperforms digital and\u0000quantum approaches by over 4 orders of magnitude in energy efficiency.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141571840","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In neural network (NN) security, safeguarding model integrity and resilience against adversarial attacks has become paramount. This study investigates the application of stochastic computing (SC) as a novel mechanism to fortify NN models. The primary objective is to assess the efficacy of SC to mitigate the deleterious impact of attacks on NN results. Through a series of rigorous experiments and evaluations, we explore the resilience of NNs employing SC when subjected to adversarial attacks. Our findings reveal that SC introduces a robust layer of defense, significantly reducing the susceptibility of networks to attack-induced alterations in their outcomes. This research contributes novel insights into the development of more secure and reliable NN systems, essential for applications in sensitive domains where data integrity is of utmost concern.
在神经网络（NN）安全方面，保障模型的完整性和抵御对抗性攻击的能力已变得至关重要。本研究调查了随机计算（SC）作为一种新机制在强化神经网络模型方面的应用。主要目的是评估随机计算在减轻攻击对 NN 结果的有害影响方面的功效。通过一系列严格的实验和评估，我们探索了采用 SC 的 NN 在受到对抗性攻击时的恢复能力。我们的研究结果表明，SC 引入了一个强大的防御层，大大降低了网络对攻击引起的结果改变的敏感性。这项研究为开发更安全、更可靠的 NN 系统提供了新的见解，这对于数据完整性最重要的敏感领域的应用至关重要。
{"title":"Late Breaking Results: Fortifying Neural Networks: Safeguarding Against Adversarial Attacks with Stochastic Computing","authors":"Faeze S. Banitaba, Sercan Aygun, M. Hassan Najafi","doi":"arxiv-2407.04861","DOIUrl":"https://doi.org/arxiv-2407.04861","url":null,"abstract":"In neural network (NN) security, safeguarding model integrity and resilience\u0000against adversarial attacks has become paramount. This study investigates the\u0000application of stochastic computing (SC) as a novel mechanism to fortify NN\u0000models. The primary objective is to assess the efficacy of SC to mitigate the\u0000deleterious impact of attacks on NN results. Through a series of rigorous\u0000experiments and evaluations, we explore the resilience of NNs employing SC when\u0000subjected to adversarial attacks. Our findings reveal that SC introduces a\u0000robust layer of defense, significantly reducing the susceptibility of networks\u0000to attack-induced alterations in their outcomes. This research contributes\u0000novel insights into the development of more secure and reliable NN systems,\u0000essential for applications in sensitive domains where data integrity is of\u0000utmost concern.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"35 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141571841","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Resistive random-access memory (RRAM) is gaining popularity due to its ability to offer computing within the memory and its non-volatile nature. The unique properties of RRAM, such as binary switching, multi-state switching, and device variations, can be leveraged to design novel techniques and algorithms. This thesis proposes a technique for utilizing RRAM devices in three major directions: i) digital logic implementation, ii) multi-valued computing, and iii) hardware security primitive design. We proposed new algorithms and architectures and conducted textit{experimental studies} on each implementation. Moreover, we developed the electronic design automation framework and hardware platforms to facilitate these experiments.
{"title":"Resistive Memory for Computing and Security: Algorithms, Architectures, and Platforms","authors":"Simranjeet Singh, Farhad Merchant, Sachin Patkar","doi":"arxiv-2407.03843","DOIUrl":"https://doi.org/arxiv-2407.03843","url":null,"abstract":"Resistive random-access memory (RRAM) is gaining popularity due to its\u0000ability to offer computing within the memory and its non-volatile nature. The\u0000unique properties of RRAM, such as binary switching, multi-state switching, and\u0000device variations, can be leveraged to design novel techniques and algorithms.\u0000This thesis proposes a technique for utilizing RRAM devices in three major\u0000directions: i) digital logic implementation, ii) multi-valued computing, and\u0000iii) hardware security primitive design. We proposed new algorithms and\u0000architectures and conducted textit{experimental studies} on each\u0000implementation. Moreover, we developed the electronic design automation\u0000framework and hardware platforms to facilitate these experiments.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"367 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141571839","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hoa T. Nguyen, Bui Binh An Pham, Muhammad Usman, Rajkumar Buyya
Quantum computing has the potential to solve complex problems beyond the capabilities of classical computers. However, its practical use is currently limited due to early-stage quantum software engineering and the constraints of Noisy Intermediate-Scale Quantum (NISQ) devices. To address this issue, this chapter introduces the concept of serverless quantum computing with examples using QFaaS, a practical Quantum Function-as-a-Service framework. This framework utilizes the serverless computing model to simplify quantum application development and deployment by abstracting the complexities of quantum hardware and enhancing application portability across different quantum software development kits and quantum backends. The chapter provides comprehensive documentation and guidelines for deploying and using QFaaS, detailing the setup, component deployment, and examples of service-oriented quantum applications. This framework offers a promising approach to overcoming current limitations and advancing the practical software engineering of quantum computing.
{"title":"Quantum Serverless Paradigm and Application Development using the QFaaS Framework","authors":"Hoa T. Nguyen, Bui Binh An Pham, Muhammad Usman, Rajkumar Buyya","doi":"arxiv-2407.02828","DOIUrl":"https://doi.org/arxiv-2407.02828","url":null,"abstract":"Quantum computing has the potential to solve complex problems beyond the\u0000capabilities of classical computers. However, its practical use is currently\u0000limited due to early-stage quantum software engineering and the constraints of\u0000Noisy Intermediate-Scale Quantum (NISQ) devices. To address this issue, this\u0000chapter introduces the concept of serverless quantum computing with examples\u0000using QFaaS, a practical Quantum Function-as-a-Service framework. This\u0000framework utilizes the serverless computing model to simplify quantum\u0000application development and deployment by abstracting the complexities of\u0000quantum hardware and enhancing application portability across different quantum\u0000software development kits and quantum backends. The chapter provides\u0000comprehensive documentation and guidelines for deploying and using QFaaS,\u0000detailing the setup, component deployment, and examples of service-oriented\u0000quantum applications. This framework offers a promising approach to overcoming\u0000current limitations and advancing the practical software engineering of quantum\u0000computing.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"21 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141546437","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the rapid advances in deep learning and smart manufacturing in Industry 4.0, there is an imperative for high-throughput, high-performance, and fully integrated visual inspection systems. Most anomaly detection approaches using defect detection datasets, such as MVTec AD, employ one-class models that require fitting separate models for each class. On the contrary, unified models eliminate the need for fitting separate models for each class and significantly reduce cost and memory requirements. Thus, in this work, we experiment with considering a unified multi-class setup. Our experimental study shows that multi-class models perform at par with one-class models for the standard MVTec AD dataset. Hence, this indicates that there may not be a need to learn separate object/class-wise models when the object classes are significantly different from each other, as is the case of the dataset considered. Furthermore, we have deployed three different unified lightweight architectures on the CPU and an edge device (NVIDIA Jetson Xavier NX). We analyze the quantized multi-class anomaly detection models in terms of latency and memory requirements for deployment on the edge device while comparing quantization-aware training (QAT) and post-training quantization (PTQ) for performance at different precision widths. In addition, we explored two different methods of calibration required in post-training scenarios and show that one of them performs notably better, highlighting its importance for unsupervised tasks. Due to quantization, the performance drop in PTQ is further compensated by QAT, which yields at par performance with the original 32-bit Floating point in two of the models considered.
{"title":"Unified Anomaly Detection methods on Edge Device using Knowledge Distillation and Quantization","authors":"Sushovan Jena, Arya Pulkit, Kajal Singh, Anoushka Banerjee, Sharad Joshi, Ananth Ganesh, Dinesh Singh, Arnav Bhavsar","doi":"arxiv-2407.02968","DOIUrl":"https://doi.org/arxiv-2407.02968","url":null,"abstract":"With the rapid advances in deep learning and smart manufacturing in Industry\u00004.0, there is an imperative for high-throughput, high-performance, and fully\u0000integrated visual inspection systems. Most anomaly detection approaches using\u0000defect detection datasets, such as MVTec AD, employ one-class models that\u0000require fitting separate models for each class. On the contrary, unified models\u0000eliminate the need for fitting separate models for each class and significantly\u0000reduce cost and memory requirements. Thus, in this work, we experiment with\u0000considering a unified multi-class setup. Our experimental study shows that\u0000multi-class models perform at par with one-class models for the standard MVTec\u0000AD dataset. Hence, this indicates that there may not be a need to learn\u0000separate object/class-wise models when the object classes are significantly\u0000different from each other, as is the case of the dataset considered.\u0000Furthermore, we have deployed three different unified lightweight architectures\u0000on the CPU and an edge device (NVIDIA Jetson Xavier NX). We analyze the\u0000quantized multi-class anomaly detection models in terms of latency and memory\u0000requirements for deployment on the edge device while comparing\u0000quantization-aware training (QAT) and post-training quantization (PTQ) for\u0000performance at different precision widths. In addition, we explored two\u0000different methods of calibration required in post-training scenarios and show\u0000that one of them performs notably better, highlighting its importance for\u0000unsupervised tasks. Due to quantization, the performance drop in PTQ is further\u0000compensated by QAT, which yields at par performance with the original 32-bit\u0000Floating point in two of the models considered.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141546438","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The quantum cloud computing paradigm presents unique challenges in task placement due to the dynamic and heterogeneous nature of quantum computation resources. Traditional heuristic approaches fall short in adapting to the rapidly evolving landscape of quantum computing. This paper proposes DRLQ, a novel Deep Reinforcement Learning (DRL)-based technique for task placement in quantum cloud computing environments, addressing the optimization of task completion time and quantum task scheduling efficiency. It leverages the Deep Q Network (DQN) architecture, enhanced with the Rainbow DQN approach, to create a dynamic task placement strategy. This approach is one of the first in the field of quantum cloud resource management, enabling adaptive learning and decision-making for quantum cloud environments and effectively optimizing task placement based on changing conditions and resource availability. We conduct extensive experiments using the QSimPy simulation toolkit to evaluate the performance of our method, demonstrating substantial improvements in task execution efficiency and a reduction in the need to reschedule quantum tasks. Our results show that utilizing the DRLQ approach for task placement can significantly reduce total quantum task completion time by 37.81% to 72.93% and prevent task rescheduling attempts compared to other heuristic approaches.
{"title":"DRLQ: A Deep Reinforcement Learning-based Task Placement for Quantum Cloud Computing","authors":"Hoa T. Nguyen, Muhammad Usman, Rajkumar Buyya","doi":"arxiv-2407.02748","DOIUrl":"https://doi.org/arxiv-2407.02748","url":null,"abstract":"The quantum cloud computing paradigm presents unique challenges in task\u0000placement due to the dynamic and heterogeneous nature of quantum computation\u0000resources. Traditional heuristic approaches fall short in adapting to the\u0000rapidly evolving landscape of quantum computing. This paper proposes DRLQ, a\u0000novel Deep Reinforcement Learning (DRL)-based technique for task placement in\u0000quantum cloud computing environments, addressing the optimization of task\u0000completion time and quantum task scheduling efficiency. It leverages the Deep Q\u0000Network (DQN) architecture, enhanced with the Rainbow DQN approach, to create a\u0000dynamic task placement strategy. This approach is one of the first in the field\u0000of quantum cloud resource management, enabling adaptive learning and\u0000decision-making for quantum cloud environments and effectively optimizing task\u0000placement based on changing conditions and resource availability. We conduct\u0000extensive experiments using the QSimPy simulation toolkit to evaluate the\u0000performance of our method, demonstrating substantial improvements in task\u0000execution efficiency and a reduction in the need to reschedule quantum tasks.\u0000Our results show that utilizing the DRLQ approach for task placement can\u0000significantly reduce total quantum task completion time by 37.81% to 72.93% and\u0000prevent task rescheduling attempts compared to other heuristic approaches.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"21 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141546337","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In-memory computing (IMC) has gained significant attention recently as it attempts to reduce the impact of memory bottlenecks. Numerous schemes for digital IMC are presented in the literature, focusing on logic operations. Often, an application's description has data dependencies that must be resolved. Contemporary IMC architectures perform read followed by write operations for this purpose, which results in performance and energy penalties. To solve this fundamental problem, this paper presents in-memory mirroring (IMM). IMM eliminates the need for read and write-back steps, thus avoiding energy and performance penalties. Instead, we perform data movement within memory, involving row-wise and column-wise data transfers. Additionally, the IMM scheme enables parallel cloning of entire row (word) with a complexity of $mathcal{O}(1)$. Moreover, our analysis of the energy consumption of the proposed technique using resistive random-access memory crossbar and experimentally validated JART VCM v1b model. The IMM increases energy efficiency and shows 2$times$ performance improvement compared to conventional data movement methods.
{"title":"In-Memory Mirroring: Cloning Without Reading","authors":"Simranjeet Singh, Ankit Bende, Chandan Kumar Jha, Vikas Rana, Rolf Drechsler, Sachin Patkar, Farhad Merchant","doi":"arxiv-2407.02921","DOIUrl":"https://doi.org/arxiv-2407.02921","url":null,"abstract":"In-memory computing (IMC) has gained significant attention recently as it\u0000attempts to reduce the impact of memory bottlenecks. Numerous schemes for\u0000digital IMC are presented in the literature, focusing on logic operations.\u0000Often, an application's description has data dependencies that must be\u0000resolved. Contemporary IMC architectures perform read followed by write\u0000operations for this purpose, which results in performance and energy penalties.\u0000To solve this fundamental problem, this paper presents in-memory mirroring\u0000(IMM). IMM eliminates the need for read and write-back steps, thus avoiding\u0000energy and performance penalties. Instead, we perform data movement within\u0000memory, involving row-wise and column-wise data transfers. Additionally, the\u0000IMM scheme enables parallel cloning of entire row (word) with a complexity of\u0000$mathcal{O}(1)$. Moreover, our analysis of the energy consumption of the\u0000proposed technique using resistive random-access memory crossbar and\u0000experimentally validated JART VCM v1b model. The IMM increases energy\u0000efficiency and shows 2$times$ performance improvement compared to conventional\u0000data movement methods.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"24 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141546436","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Decentralized Intelligence Network (DIN) addresses the significant challenges of data sovereignty and AI utilization caused by the fragmentation and siloing of data across providers and institutions. This comprehensive framework overcomes access barriers to scalable data sources previously hindered by silos by leveraging: 1) personal data stores as a prerequisite for data sovereignty; 2) a scalable federated learning protocol implemented on a public blockchain for decentralized AI training, where data remains with participants and only model parameter updates are shared; and 3) a scalable, trustless rewards mechanism to incentivize participation and ensure fair reward distribution. This framework ensures that no entity can prevent or control access to training on data offered by participants or determine financial benefits, as these processes operate on a public blockchain with an immutable record and without a third party. It supports effective AI training, allowing participants to maintain control over their data, benefit financially, and contribute to a decentralized, scalable ecosystem that leverages collective AI to develop beneficial algorithms.
{"title":"Decentralized Intelligence Network (DIN)","authors":"Abraham Nash","doi":"arxiv-2407.02461","DOIUrl":"https://doi.org/arxiv-2407.02461","url":null,"abstract":"Decentralized Intelligence Network (DIN) addresses the significant challenges\u0000of data sovereignty and AI utilization caused by the fragmentation and siloing\u0000of data across providers and institutions. This comprehensive framework\u0000overcomes access barriers to scalable data sources previously hindered by silos\u0000by leveraging: 1) personal data stores as a prerequisite for data sovereignty;\u00002) a scalable federated learning protocol implemented on a public blockchain\u0000for decentralized AI training, where data remains with participants and only\u0000model parameter updates are shared; and 3) a scalable, trustless rewards\u0000mechanism to incentivize participation and ensure fair reward distribution.\u0000This framework ensures that no entity can prevent or control access to training\u0000on data offered by participants or determine financial benefits, as these\u0000processes operate on a public blockchain with an immutable record and without a\u0000third party. It supports effective AI training, allowing participants to\u0000maintain control over their data, benefit financially, and contribute to a\u0000decentralized, scalable ecosystem that leverages collective AI to develop\u0000beneficial algorithms.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"137 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141522518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
While supervised learning has achieved significant success in computer vision tasks, acquiring high-quality annotated data remains a bottleneck. This paper explores both scholarly and non-scholarly works in AI-assistive deep learning image annotation systems that provide textual suggestions, captions, or descriptions of the input image to the annotator. This potentially results in higher annotation efficiency and quality. Our exploration covers annotation for a range of computer vision tasks including image classification, object detection, regression, instance, semantic segmentation, and pose estimation. We review various datasets and how they contribute to the training and evaluation of AI-assistive annotation systems. We also examine methods leveraging neuro-symbolic learning, deep active learning, and self-supervised learning algorithms that enable semantic image understanding and generate free-text output. These include image captioning, visual question answering, and multi-modal reasoning. Despite the promising potential, there is limited publicly available work on AI-assistive image annotation with textual output capabilities. We conclude by suggesting future research directions to advance this field, emphasizing the need for more publicly accessible datasets and collaborative efforts between academia and industry.
{"title":"Assistive Image Annotation Systems with Deep Learning and Natural Language Capabilities: A Review","authors":"Moseli Mots'oehli","doi":"arxiv-2407.00252","DOIUrl":"https://doi.org/arxiv-2407.00252","url":null,"abstract":"While supervised learning has achieved significant success in computer vision\u0000tasks, acquiring high-quality annotated data remains a bottleneck. This paper\u0000explores both scholarly and non-scholarly works in AI-assistive deep learning\u0000image annotation systems that provide textual suggestions, captions, or\u0000descriptions of the input image to the annotator. This potentially results in\u0000higher annotation efficiency and quality. Our exploration covers annotation for\u0000a range of computer vision tasks including image classification, object\u0000detection, regression, instance, semantic segmentation, and pose estimation. We\u0000review various datasets and how they contribute to the training and evaluation\u0000of AI-assistive annotation systems. We also examine methods leveraging\u0000neuro-symbolic learning, deep active learning, and self-supervised learning\u0000algorithms that enable semantic image understanding and generate free-text\u0000output. These include image captioning, visual question answering, and\u0000multi-modal reasoning. Despite the promising potential, there is limited\u0000publicly available work on AI-assistive image annotation with textual output\u0000capabilities. We conclude by suggesting future research directions to advance\u0000this field, emphasizing the need for more publicly accessible datasets and\u0000collaborative efforts between academia and industry.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"5 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141508407","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：