SecMDGM: Federated Learning Security Mechanism Based on Multi−Dimensional Auctions

Abstract

As a newly emerging distributed machine learning technology, federated learning has unique advantages in the era of big data. We explore how to motivate participants to experience auctions more actively and safely. It is also essential to ensure that the final participant who wins the right to participate can guarantee relatively high−quality data or computational performance. Therefore, a secure, necessary and effective mechanism is needed through strict theoretical proof and experimental verification. The traditional auction theory is mainly oriented to price, not giving quality issues as much consideration. Hence, it is challenging to discover the optimal mechanism and solve the privacy problem when considering multi−dimensional auctions. Therefore, we (1) propose a multi−dimensional information security mechanism, (2) propose an optimal mechanism that satisfies the Pareto optimality and incentive compatibility named the SecMDGM and (3) verify that for the aggregation model based on vertical data, this mechanism can improve the performance by 2.73 times compared to that of random selection. These are all important, and they complement each other instead of being independent or in tandem. Due to security issues, it can be ensured that the optimal multi−dimensional auction has practical significance and can be used in verification experiments.

A Non-Negative Matrix Factorization Framework for Privacy-Preserving and Federated Learning

“…This precision in identifying anomalies was crucial for maintaining the integrity of the FL model, particularly in scenarios where data reliability is paramount. Our approach maintained commendable performance when transitioning to CIFAR-10, a dataset characterized by its higher complexity and variability, as shown in Figure (3). Despite the increased difficulty in discerning anomalies amidst more complex data patterns, the model adeptly adjusted, showcasing its robustness and adaptability.…”

“…These susceptibilities underscore an urgent need for robust defense mechanisms capable of safeguarding the federated model's integrity without infringing upon the privacy of the participant entities. The complexity of securing FL is magnified by the distributed nature of its architecture, which introduces unique challenges such as data heterogeneity, communication inefficiencies, and the risk of isolation attacks [3], [4]. The quest for a resilient FL framework is technical and strategic, necessitating solutions that are as dynamic and adaptable as the threats they contend with.…”

Revolutionizing AI-Assisted Education with Federated Learning: A Pathway to Distributed, Privacy-Preserving, and Debiased Learning Ecosystems

“…Multi-Modal FL over Unbalanced Modalities: Research in domains like medicine and wireless networks have demonstrated the potential of multi-modal FL, which enables using different types/modalities of data during ML model training (Xiong et al 2022;Lin et al 2023;Che et al 2023;Borazjani et al 2024). Similarly, educational data, which encompasses a variety of modalities, e.g., text, audio, video, and interactive activities, is inherently multi-modal.…”

Multi-Robot Privacy-Preserving Algorithms Based on Federated Learning: A Review

“…The authors of the literature [56] proposed a privacy-enhanced FL scheme in industrial robots for combating collusion attacks. The authors in the literature [57] combined multi-robot auction algorithms with FL, where the default server is trusted in the article, and privacy-preserving bidding, auction computation, and model aggregation processes are performed using partial HE.…”