检索结果-内蒙古大学图书馆

Trustworthy distributed ai Systems: Robustness, Privacy, and Governance

ACM COMPUTING SURVEYS 2025年第6期57卷 1-42页

作者： Wei, Wenqi Liu, Ling Fordham Univ 113 West 60th StLL610H New York NY 10023 USA Georgia Inst Technol 226 Ferst DrKACB room 3340 Atlanta GA USA

Emerging distributed ai systems are revolutionizing big data computing and data processing capabilities with growing economic and societal impact. However, recent studies have identified new attack surfaces and risks caused by security, privacy, and fairness issues in ai systems. In this article, we review representative techniques, algorithms, and theoretical foundations for trustworthy distributed ai through robustness guarantee, privacy protection, and fairness awareness in distributed learning. We first provide a brief overview of alternative architectures for distributed learning, discuss inherent vulnerabilities for security, privacy, and fairness of ai algorithms in distributed learning, and analyze why these problems are present in distributed learning regardless of specific architectures. Then, we provide a unique taxonomy of countermeasures for trustworthy distributed ai, covering (1) robustness to evasion attacks and irregular queries at inference, and robustness to poisoning attacks, Byzantine attacks, and irregular data distribution during training;(2) privacy protection during distributed learning and model inference at deployment;and (3) ai fairness and governance with respect to both data and models. We conclude with a discussion on open challenges and future research directions toward trustworthy distributed ai, such as the need for trustworthy ai policy guidelines, the ai responsibility-utility co-design, and incentives and compliance.

关键词： distributed ai trustworhty ai ai Governance

来源：评论

学校读者我要写书评

暂无评论

From Technical Prerequisites to Improved Care: distributed Edge ai for Tomographic Imaging

引用

IEEE ACCESS 2025年 13卷 14317-14343页

作者： Akdemir, Bilgehan Shahid, Hafiz Faheem Brix, Mikael A. K. Laakkola, Juho Islam, Johirul Kumar, Tanesh Reponen, Jarmo Nieminen, Miika T. Harjula, Erkki Univ Oulu Ctr Wireless Commun Networks & Syst CWC NS Oulu 90570 Finland Univ Oulu Res Unit Hlth Sci & Technol HST Oulu 90220 Finland Oulu Univ Hosp Dept Diagnost Radiol Oulu 90220 Finland Aalto Univ Dept Informat & Commun Engn Espoo 02150 Finland

Recent years have seen a surge in ai-driven medical image processing, leading to significant improvements in diagnostic performance. However, medical imaging technologies tend to create staggering volumes of medical data, necessitating high-performance computing. Cloud systems with robust GPUs and resource capacity are optimal choices for DL-based medical image processing. However, transferring data to the cloud for processing strains communication links, introduces high communication latency, and raises privacy and security concerns. Consequently, despite the undisputed benefits of cloud computing, dedicated standalone local computers are still used for image reconstruction in today's systems. This localized strategy uses expensive hardware inefficiently and falls short of scalability and maintainability. Edge computing emerges as an innovative concept by bringing cloud processing capabilities closer to data sources. A continuum of computing including local, edge, and cloud tiers would offer a promising solution for medical image processing. According to literature survey, there are no significant works on utilizing edge cloud continuum for CBCT imaging. To fill this gap, we introduce novel 3-TECC architectural concept, specifically designed for CBCT data reconstruction in medical imaging. This article explores the evolving synergy among medical imaging, distributed ai, containerized solutions, and edge-cloud continuum technologies, highlighting their clinical implications and illuminating the potential for transformative patient care. We uncover challenges and opportunities this convergence provides with the CBCT image reconstruction use case, while aligning with regulatory compliance. The proposed 3-TECC architecture advocates a decentralized data processing paradigm, reducing reliance on the centralized approach and emphasizing the role of local-edge computing.

关键词： Imaging Medical diagnostic imaging Cloud computing Image reconstruction Image edge detection Computer architecture Computed tomography Edge ai Hospitals Technical requirements CBCT distributed ai edge computing edge cloud continuum GDPR medical imaging

来源：评论

学校读者我要写书评

暂无评论

distributed ai for ambient intelligence: Issues and approaches

Distributed AI for ambient intelligence: Issues and approach...

引用

European Conference on Ambient Intelligence

作者： Patkos, Theodore Bikakis, Antonis Antoniou, Grigoris Papadopouli, Maria Plexousakis, Dimitris FORTH Vassilika Vouton Inst Comp Sci GR-71110 Iraklion Greece

ISBN: (纸本)9783540766513

Research in many fields of ai, such as distributed planning and reasoning, agent teamwork and coalition formation, cooperative problem solving and action theory has advanced significantly over the last years, both from a theoretical and a practical perspective. In the light of the development towards ambient, pervasive and ubiquitous computing, this research will be tested under new, more demanding realistic conditions, stimulating the emergence of novel approaches to handle the challenges that these open, dynamic environments introduce. This paper identifies shortcomings of state-of-the-art techniques in handling the complexity of the Ambient Intelligence vision, motivated by the experience gained during the development and usage of a context-aware platform for mobile devices in dynamic environments. The paper raises research issues and discusses promising directions for realizing the objectives of near-future pervasive information systems.

关键词： ambient intelligence distributed ai context awareness action theories multi-agent cooperation

来源：评论

学校读者我要写书评

暂无评论

A distributed ai Framework for Nano-Grid Power Management and Control

引用

IEEE ACCESS 2024年 12卷 43350-43377页

作者： Ioannou, Iacovos I. Javaid, Saher Christophorou, Christophoros Vassiliou, Vasos Pitsillides, Andreas Tan, Yasuo Univ Cyprus Dept Comp Sci CY-1678 Nicosia Cyprus CYENS Ctr Excellence CY-1016 Nicosia Cyprus Japan Adv Inst Sci & Technol Sch Informat Sci Nomi Ishikawa 9231292 Japan Univ Johannesburg Dept Elect & Elect Engn Sci ZA-2092 Johannesburg South Africa

Due to their minimal environmental impact, green energy sources like wind turbines and solar panels are increasingly utilized in power systems. However, the power they generate is highly variable, leading to unpredictable fluctuations in power supply. Additionally, advanced smart functions in consumer devices and their unpredictable usage patterns contribute to similar fluctuations in power consumption. These fluctuations present a significant challenge to the stability and quality of the power grid, creating a complex issue of power imbalance that becomes harder to manage. Innovative management and control approaches are necessary to address these challenges and thus support the shift to sustainable energy sources. Artificial intelligence (ai) techniques are increasingly proposed as promising solutions, albeit mostly implemented as isolated solutions within centralized power control systems. To effectively manage the complex and often large scale power systems, this paper advocates the use of a distributed ai (Dai) framework as imperative in enhancing their agility and stability. An illustrative Nano-Grid example (including the potential use of battery sources in extreme scenarios) is adopted to demonstrate the framework's utility, and a number of power control strategies to safeguard the power system against the variability of both power generators and loads are theoretically formulated and then realized within the proposed framework. Linear Programming, Ant Colony Optimization, Genetic Algorithms, and Particle Swarm Optimization techniques are experimented with, and through simulations, the utility of the Dai framework is demonstrated. The findings underscore the effectiveness and potential benefits of the proposed framework in ensuring the safe and effective operation of power systems with the use of particle swarm optimization amid fluctuating energy scenarios with a small to large number of devices in the nano-grid.

关键词： distributed ai power fluctuations distributed power sources distributed power loads power balance power control

来源：评论

学校读者我要写书评

暂无评论

Supporting Architectural Design Process with FLEA A distributed ai Methodology for Retrieval, Suggestion, Adaptation, and Explanation of Room Configurations 18th

Supporting Architectural Design Process with FLEA A Distribu...

引用

18th International Conference on Computer-aided Architectural Design (CAAD Futures) - Hello, Culture

作者： Eisenstadt, Viktor Lanhgenhan, Christoph Althoff, Klaus-Dieter Univ Hildesheim Inst Comp Sci Samelsonpl 1 D-31141 Hildesheim Germany Tech Univ Munich Chair Architectural Informat Arcisstr 21 D-80333 Munich Germany German Res Ctr Artificial Intelligence DFKI Trippstadter Str 122 D-67663 Kaiserslautern Germany

ISBN: (纸本)9789811384103;9789811384097

The artificial intelligence methods, such as case-based reasoning and artificial neural networks were already applied to the task of architectural design support in a multitude of specific approaches and tools. However, modern ai trends, such as Explainable ai (Xai), and additional features, such as providing contextual suggestions for the next step of the design process, were rarely considered an integral part of these approaches or simply not available. In this paper, we present an application of a distributed ai-based methodology FLEA (Find, Learn, Explain, Adapt) to the task of room configuration during the early conceptual phases of architectural design. The implementation of the methodology in the framework MetisCBR applies CBR-based methods for retrieval of similar floor plans to suggest possibly inspirational designs and to explain the returned results with specific explanation patterns. Furthermore, it makes use of a farm of recurrent neural networks to suggest contextually suitable next configuration steps and to present design variations that show how the designs may evolve in the future. The flexibility of FLEA allows for variational use of its components in order to activate the currently required modules only. The methodology was initialized during the basic research project Metis (funded by German Research Foundation) during which the architectural semantic search patterns and a family of corresponding floor plan representations were developed. FLEA uses these patterns and representations as the base for its semantic search, explanation, next step suggestion, and adaptation components. The methodology implementation was iteratively tested during quantitative evaluations and user studies with multiple floor plan datasets.

关键词： Room configuration distributed ai Case-based reasoning Neural networks Explainable ai

来源：评论

学校读者我要写书评

暂无评论

Applying an Agent-based distributed ai Framework to Forecast Power for the Mini-Grid Stability 11

Applying an Agent-based Distributed AI Framework to Forecast...

引用

11th IEEE International Conference on Consumer Electronics - Taiwan (ICCE-Taiwan) - Empower of Innovative Consumer Technology

作者： Ioannou, Iacovos I. Javaid, Saher Vassiliou, Vasos Pitsillides, Andreas Tan, Yasuo Japan Adv Inst Sci & Technol Sch Informat Sci Nomi Ishikawa Japan Univ Cyprus Nicosia Cyprus CYENS Ctr Excellence Nicosia Cyprus

ISBN: (纸本)9798350386851;9798350386844

Renewable energy sources, expected to form about 70% of power systems by 2050, bring challenges like fluctuating outputs and grid instability. Advanced power monitoring systems, crucial in environments like research facilities and hospitals, must navigate these dynamic scenarios. Traditional power management, especially Uninterruptible Power Supply (UPS) systems, often needs to catch up due to high costs and limited response to varied power demands, focusing mainly on constant power supply without differentiating between constant and fluctuating loads. In response, Artificial intelligence (ai) techniques are becoming indispensable for real-time power prediction and control. A distributed ai framework forecasts power needs, considering renewable sources, loads, and storage. This is key to ensuring smooth mini-grid operations, balancing operational demands with environmental considerations, and advancing intelligent energy management. Such systems are essential in optimizing energy usage, aligning it with available power to enhance efficiency and reduce waste. This is particularly important for mini-grids, with or without UPS systems, where predictive monitoring can substantially cut operational costs and extend lifespan. The paper focuses on providing consistent, constant, and fluctuating power by predicting mini-grid power needs hourly from the previous day's data. We use a Temporal Convolutional Network (TCN) for time series prediction, integrated within the BDIx agent's belief system through TensorFlow Lite. This approach accurately predicts upcoming power needs, ensures smooth operation, and prevents power outages. The TCN model's predictive capabilities highlight a significant stride in combining ai with energy management to address the complexities of modern power systems.

关键词： distributed ai power fluctuations distributed power sources distributed power loads constant power load prediction fluctuating power prediction power control

来源：评论

学校读者我要写书评

暂无评论

Interplay between distributed ai Workflow and URLLC

Interplay between Distributed AI Workflow and URLLC

引用

IEEE Global Communications Conference (GLOBECOM)

作者： Ganjalizadeh, Milad Ghadikolaei, Hossein S. Haraldson, Johan Petrova, Marina Ericsson Res Stockholm Sweden KTH Royal Inst Technol Sch Elect Engn & Comp Sci Stockholm Sweden Rhein Westfal TH Aachen Mobile Commun & Comp Aachen Germany

ISBN: (纸本)9781665435406

distributed artificial intelligence (ai) has recently accomplished tremendous breakthroughs in various communication services, ranging from fault-tolerant factory automation to smart cities. When distributed learning is run over a set of wireless connected devices, random channel fluctuations, and the incumbent services simultaneously running on the same network affect the performance of distributed learning. In this paper, we investigate the interplay between distributed ai workflow and ultra-reliable low latency communication (URLLC) services running concurrently over a network. Using 3GPP compliant simulations in a factory automation use case, we show the impact of various distributed ai settings (e.g., model size and the number of participating devices) on the convergence time of distributed ai and the application layer performance of URLLC. Unless we leverage the existing 5G-NR quality of service handling mechanisms to separate the traffic from the two services, our simulation results show that the impact of distributed ai on the availability of the URLLC devices is significant. Moreover, with proper setting of distributed ai (e.g., proper user selection), we can substantially reduce network resource utilization, leading to lower latency for distributed ai and higher availability for the URLLC users. Our results provide important insights for future 6G and ai standardization.

关键词： 6G availability distributed ai factory automation federated learning quality-of-service URLLC

来源：评论

学校读者我要写书评

暂无评论

Policy Enabled Caching for distributed ai 5

Policy Enabled Caching for Distributed AI

引用

IEEE International Conference on Big Data (IEEE Big Data)

作者： Verma, Dinesh C. Bent, Graham IBM TJ Watson Res Ctr Yorktown Hts NY 10598 USA IBM Res UK Warrington Cheshire England

ISBN: (纸本)9781538627150

Web Caching has established itself as a key enabling technology within the Internet. It enables efficient browsing of websites and web-based services on networks that are bandwidth constrained. However, similar techniques are not available for ai based solutions. Many ai solutions are based on deep neural networks or similar approaches which require creation of machine learning models trained with huge amounts of data. Such models are best created in centralized locations with significant processing power. In many environments, sending the data to a centralized location is infeasible or undesirable. A judicious combination of ideas borrowed from web-caching paradigm, with ideas from ai and machine learning can provide an effective solution for exploitation of deep learning models in bandwidth constrained environments. Allowing such caches to generate their own policies using a generative policy approach can enable the creation of a generic edge caching system which can be used with a wide variety of backend ai systems.

关键词： Edge Computing Fog Computing distributed ai Semantic Caching Generative Policy

来源：评论

学校读者我要写书评

暂无评论

An Initial Analysis of the Shortcomings of Conventional ai and the Benefits of distributed ai Approaches in Industrial Use Cases 17th

An Initial Analysis of the Shortcomings of Conventional AI a...

引用

17th IFIP WG 12.5 International Conference on Artificial Intelligence Applications and Innovations (aiai) / 6th 5G-PINE Workshop / ai-BIO Workshop / DAai Workshop / DARE Workshop / EEai Workshop / 10th MHDW Workshop

作者： Hristoskova, Anna Gonzalez-Deleito, Nicolas Klein, Sarah Sousa, Joana Martins, Nuno Tagaio, Joao Serra, Joao Silva, Carlos Ferreira, Joao Santos, Pedro M. Morla, Ricardo Almeida, Luis Bulut, Baris Sultanoglu, Sencer Sirris Brussels Belgium NOS Inovacao Lisbon Portugal Bold Int Lisbon Portugal Caixa Mag Lisbon Portugal KCSIT Lisbon Portugal Inst Super Engn Porto Porto Portugal Univ Porto Fac Engn Porto Portugal CISTER Res Ctr Real Time & Embedded Comp Syst Porto Portugal Enforma Bilisim Istanbul Turkey Eliar Istanbul Turkey

ISBN: (纸本)9783030791599;9783030791575;9783030791568

The centralised approach of IoT (Internet of Things) applications leveraging cloud infrastructures to address constraints at the level of end and edge nodes is no longer viable, especially for applications with hard real-time requirements and increasing ai (Artificial Intelligence) usage. This paper presents an initial analysis of the shortcomings of such centralised ai approaches applied to the five industrial use cases considered in the ITEA3 MIRai project and discusses the expected benefits that distributed ai approaches will bring to these use cases, namely, to lift constraints such as computing power, bandwidth, latency, security and privacy.

关键词： Artificial Intelligence (ai) distributed ai Secure analytics Access control Privacy

来源：评论

学校读者我要写书评

暂无评论

Real Time Supervision of Wastewater Treatment Plants: A distributed ai Approach

引用

IFAC Proceedings Volumes 1994年第10期27卷 188-193页

作者： M. Sànchez L. R-Roda J. Lafuente U. Cortés M. Poch Universitat Politiénica de Catalunya. Department of Software. Pau Garga1lo 5. 08028 Barcelona Universitat Autónoma de Barcelona. Chemical Engineering Unit. 08193 Bellaterra Spain

Conventional automatic control systems applied to wastewater treatment plants deals with some difficulties: complexity of the system, an ill-structured domain, qualitative information, uncertainty or approximate knowledge, real-time dynamic system, etc. Also, the knowledge-based approaches developed from artificial intelligence field show other disadvantages: brittleness, knowledge acquisition problem, do not learn from experience, the increasing complexity of the systems, etc. The main goal of the paper is to present a knowledge-based distributed archi1ecture for real time supervision and control of wastewater treatment plants that overcomes some of these troubles. It is discussed the development of the application and the methodology employed in it. The prototype's architecture being developed Dai-DEPUR is detailed together with some obtained results.

关键词： Real-Time Supervision and Control distributed ai Knowledge-based Systems Wastewater treatment Knowledge Acquisition Environmental Engineering Biotechnology

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：