A Robust and Privacy-Aware Federated Learning Framework for Non-Intrusive Load Monitoring DOI
Vidushi Agarwal, Omid Ardakanian, Sujata Pal

и другие.

IEEE Transactions on Sustainable Computing, Год журнала: 2024, Номер 9(5), С. 766 - 777

Опубликована: Фев. 28, 2024

With the rollout of smart meters, a vast amount energy time-series became available from homes, enabling applications such as non-intrusive load monitoring (NILM). The inconspicuous collection this data, however, poses risk to privacy customers. Federated Learning (FL) eliminates problem sharing raw data with cloud service provider by allowing machine learning models be trained in collaborative fashion on decentralized data. Although several NILM techniques that rely FL train deep neural network for identifying consumption individual appliances have been proposed recent years, robustness these malicious users and their ability fully protect user remain unexplored. In paper, we present robust privacy-preserving FL-based framework bidirectional transformer architecture NILM. This takes advantage meta-learning algorithm handle heterogeneity prevalent real-world settings. efficacy is corroborated through comparative experiments using two datasets. results show can attain an accuracy par centrally-trained disaggregation model, while preserving privacy.

Язык: Английский

A Survey on ChatGPT: AI–Generated Contents, Challenges, and Solutions DOI Creative Commons
Yuntao Wang, Yanghe Pan, Miao Yan

и другие.

IEEE Open Journal of the Computer Society, Год журнала: 2023, Номер 4, С. 280 - 302

Опубликована: Янв. 1, 2023

With the widespread use of large artificial intelligence (AI) models such as ChatGPT, AI-generated content (AIGC) has garnered increasing attention and is leading a paradigm shift in creation knowledge representation. AIGC uses generative AI algorithms to assist or replace humans creating massive, high-quality, human-like at faster pace lower cost, based on user-provided prompts. Despite recent significant progress AIGC, security, privacy, ethical, legal challenges still need be addressed. This paper presents an in-depth survey working principles, security privacy threats, state-of-the-art solutions, future paradigm. Specifically, we first explore enabling technologies, general architecture discuss its modes key characteristics. Then, investigate taxonomy threats highlight ethical societal implications GPT technologies. Furthermore, review watermarking approaches for regulatable paradigms regarding model produced content. Finally, identify open research directions related AIGC.

Язык: Английский

Процитировано

143
A Survey on Digital Twins: Architecture, Enabling Technologies, Security and Privacy, and Future Prospects DOI
Yuntao Wang, Zhou Su, Shaolong Guo

и другие.

IEEE Internet of Things Journal, Год журнала: 2023, Номер 10(17), С. 14965 - 14987

Опубликована: Апрель 3, 2023

By interacting, synchronizing, and cooperating with its physical counterpart in real time, digital twin (DT) is promised to promote an intelligent, predictive, optimized modern city. Via interconnecting massive entities their virtual twins inter-twin intra-twin communications, the Internet of DTs (IoDT) enables free data exchange, dynamic mission cooperation, efficient information aggregation for composite insights across vast physical/virtual entities. However, as IoDT incorporates various cutting-edge technologies spawn new ecology, severe known/unknown security flaws, privacy invasions hinder wide deployment. Besides, intrinsic characteristics IoDT, such decentralized structure, information-centric routing, semantic entail critical challenges service provisioning IoDT. To this end, article presents in-depth review respect system architecture, enabling technologies, security/privacy issues. Specifically, we first explore a novel distributed architecture cyber–physical interactions discuss key communication modes. Afterward, investigate taxonomy threats research challenges, state-of-the-art defense approaches. Finally, point out trends open directions related

Язык: Английский

Процитировано

110
Unleashing the Power of Edge-Cloud Generative AI in Mobile Networks: A Survey of AIGC Services DOI
Minrui Xu, Hongyang Du, Dusit Niyato

и другие.

IEEE Communications Surveys & Tutorials, Год журнала: 2024, Номер 26(2), С. 1127 - 1170

Опубликована: Янв. 1, 2024

Artificial Intelligence-Generated Content (AIGC) is an automated method for generating, manipulating, and modifying valuable diverse data using AI algorithms creatively. This survey paper focuses on the deployment of AIGC applications, e.g., ChatGPT Dall-E, at mobile edge networks, namely that provide personalized customized services in real time while maintaining user privacy. We begin by introducing background fundamentals generative models lifecycle which includes collection, training, fine-tuning, inference, product management. then discuss collaborative cloud-edge-mobile infrastructure technologies required to support enable users access networks. Furthermore, we explore AIGC-driven creative applications use cases Additionally, implementation, security, privacy challenges deploying Finally, highlight some future research directions open issues full realization

Язык: Английский

Процитировано

98
Wild Patterns Reloaded: A Survey of Machine Learning Security against Training Data Poisoning DOI Open Access
Antonio Emanuele Cinà, Kathrin Grosse, Ambra Demontis

и другие.

ACM Computing Surveys, Год журнала: 2023, Номер 55(13s), С. 1 - 39

Опубликована: Март 1, 2023

The success of machine learning is fueled by the increasing availability computing power and large training datasets. data used to learn new models or update existing ones, assuming that it sufficiently representative will be encountered at test time. This assumption challenged threat poisoning, an attack manipulates compromise model’s performance Although poisoning has been acknowledged as a relevant in industry applications, variety different attacks defenses have proposed so far, complete systematization critical review field still missing. In this survey, we provide comprehensive learning, reviewing more than 100 papers published past 15 years. We start categorizing current then organize accordingly. While focus mostly on computer-vision argue our also encompasses state-of-the-art for other modalities. Finally, discuss resources research shed light limitations open questions field.

Язык: Английский

Процитировано

68
When Federated Learning Meets Privacy-Preserving Computation DOI
Jingxue Chen, Hang Yan,

Z. Liu

и другие.

ACM Computing Surveys, Год журнала: 2024, Номер 56(12), С. 1 - 36

Опубликована: Июль 22, 2024

Nowadays, with the development of artificial intelligence (AI), privacy issues attract wide attention from society and individuals. It is desirable to make data available but invisible, i.e., realize analysis calculation without disclosing unauthorized entities. Federated learning (FL) has emerged as a promising privacy-preserving computation method for AI. However, new have arisen in FL-based application, because various inference attacks can still infer relevant information about raw local models or gradients. This will directly lead disclosure. Therefore, it critical resist these achieve complete computation. In light overwhelming variety multitude protocols, we survey protocols series perspectives supply better comprehension researchers scholars. Concretely, classification discussed, including four kinds well malicious server poisoning attack. Besides, this article systematically captures state-of-the-art by analyzing design rationale, reproducing experiment classic schemes, evaluating all discussed terms efficiency security properties. Finally, identifies number interesting future directions.

Язык: Английский

Процитировано

47
Adversarial attacks and defenses in explainable artificial intelligence: A survey DOI
Hubert Baniecki, Przemysław Biecek

Information Fusion, Год журнала: 2024, Номер 107, С. 102303 - 102303

Опубликована: Фев. 19, 2024

Язык: Английский

Процитировано

46
Security and Privacy Challenges of AIGC in Metaverse: A Comprehensive Survey DOI
Shoulong Zhang,

Haomin Li,

Kaiwen Sun

и другие.

ACM Computing Surveys, Год журнала: 2025, Номер unknown

Опубликована: Апрель 18, 2025

The Metaverse is a hybrid environment that integrates both physical and virtual realms. has been accessible due to many facilitating technologies. One of the essential technologies contribute AIGC. It crucial in creating artificial assets presenting natural interactions efficiently effectively. Nevertheless, AIGC models encounter external internal obstacles security, privacy, ethics during every level their development. To conduct thorough analysis investigation risks threats, we propose new taxonomy system categorizes issues based on three primary factors: stage threat exposure, specific area concerns, origin threats. Furthermore, present unresolved questions prompt additional into posed by steps taken counteract them art creation interactive methodologies. This evaluation offers broad perspective security measures uses Metaverse.

Язык: Английский

Процитировано

3
Backdoor attacks and defenses in federated learning: Survey, challenges and future research directions DOI Open Access
Thuy Dung Nguyen, Tuan Nguyen, Phi Le Nguyen

и другие.

Engineering Applications of Artificial Intelligence, Год журнала: 2023, Номер 127, С. 107166 - 107166

Опубликована: Окт. 5, 2023

Язык: Английский

Процитировано

43
AIsmosis and the pas de deux of human-AI interaction: Exploring the communicative dance between society and artificial intelligence DOI Open Access
Ayşe Aslı Bozdağ

Online Journal of Communication and Media Technologies, Год журнала: 2023, Номер 13(4), С. e202340 - e202340

Опубликована: Июнь 19, 2023

As the global influence of artificial intelligence (AI) in our daily lives and looming advent general (AGI) become increasingly apparent, need for a sophisticated interpretive framework intensifies. This paper introduces ‘AIsmosis’–a term that captures AI’s gradual, nuanced integration into society, akin to biological process osmosis. dynamics are examined through lens three pivotal theories: social construction technology, technological determinism, diffusion innovations. These theories collectively elucidate sociocultural influences on AI, potential repercussions unchecked growth, factors driving adoption novel technologies. Building upon these explorations, ‘controlled AIsmosis’ conceptual emerges, emphasizing ethically conscious development, active stakeholder communication, democratic dialogue context AI technology adoption. Rooted communicative action theory, this illuminates transformative impact society. It calls comprehensive evaluation systems steer their impacts, acknowledging pervasive transcending traditional disciplinary boundaries. work underscores multidisciplinary interdisciplinary approach investigating complex AI-society interplay understanding ethical societal consequences AIsmosis.

Язык: Английский

Процитировано

20
Data-Agnostic Model Poisoning Against Federated Learning: A Graph Autoencoder Approach DOI
Kai Li, Jingjing Zheng, Xin Yuan

и другие.

IEEE Transactions on Information Forensics and Security, Год журнала: 2024, Номер 19, С. 3465 - 3480

Опубликована: Янв. 1, 2024

This paper proposes a novel, data-agnostic, model poisoning attack on Federated Learning (FL), by designing new adversarial graph autoencoder (GAE)-based framework. The requires no knowledge of FL training data and achieves both effectiveness undetectability. By listening to the benign local models global model, attacker extracts structural correlations among features substantiating models. then adversarially regenerates while maximizing loss, subsequently generates malicious using structure ones. A algorithm is designed iteratively train GAE sub-gradient descent. convergence under rigorously proved, with considerably large optimality gap. Experiments show that accuracy drops gradually proposed existing defense mechanisms fail detect it. can give rise an infection across all devices, making it serious threat FL.

Язык: Английский

Процитировано

8