Dynamic Asynchronous Anti Poisoning Federated Deep Learning with Blockchain-Based Reputation-Aware Solutions

Abstract

As promising privacy-preserving machine learning technology, federated learning enables multiple clients to train the joint global model via sharing model parameters. However, inefficiency and vulnerability to poisoning attacks significantly reduce federated learning performance. To solve the aforementioned issues, we propose a dynamic asynchronous anti poisoning federated deep learning framework to pursue both efficiency and security. This paper proposes a lightweight dynamic asynchronous algorithm considering the averaging frequency control and parameter selection for federated learning to speed up model averaging and improve efficiency, which enables federated learning to adaptively remove the stragglers with low computing power, bad channel conditions, or anomalous parameters. In addition, a novel local reliability mutual evaluation mechanism is presented to enhance the security of poisoning attacks, which enables federated learning to detect the anomalous parameter of poisoning attacks and adjust the weight proportion of in model aggregation based on evaluation score. The experiment results on three datasets illustrate that our design can reduce the training time by 30% and is robust to the representative poisoning attacks significantly, confirming the applicability of our scheme.

WITHDRAWN: Integration of Blockchain Technology with Deep Learning Approaches: Vision and Future Opportunities

“…Additionally, it guarantees that the documents were not altered during the human-assisted auditing phase. [71][72][73][74][75] Combined Decisions Private/Public The robots will get assistance from a voting-based approach for making an effective decision on the gathered information by utilizing swarm robotics on blockchain technology [76][77][78][79] Enhanced Robustness Private/Public Blockchain and deep learning integration can be beneficial in commercial environments where the parties can cooperate in a trustless and automated manner.…”

Trusted Data Storage Architecture for National Infrastructure

“…Federal learning is the local training and analysis of common data, and the selective display of training results, on the premise that multiple participants guarantee that their private data will not be leaked [ 21 , 22 , 23 , 24 , 25 ]. Federal learning can model data usage and machine learning under the requirements of user privacy protection, data security and government, which can effectively solve the problem of data islands.…”

Exploring Deep Federated Learning for the Internet of Things: A GDPR-Compliant Architecture

“…The architecture [33] accomplished collaborative fairness via local dependability, participation level, and transaction points, and the concept of different datasets is incorporated for privacy throughout the review process. The proposed technique, as per experimental data, proved resilient against poisoning attacks, leveraging networked devices for malware defense and threat classification [34].…”

“…The [33] study, proposed a FL system with concerns for fairness and robustness against poisoning assaults. By assigning the job to clients for cross-validation, a distinct local reliability mutual assessment technique is presented for identifying anomalous updates without having access to raw data.…”