Blockchain-enabled Trustworthy Federated Unlearning

TL;DR

The paper tackles the challenge of data ownership and the right to be forgotten in federated learning by proposing a blockchain-enabled trustworthy federated unlearning framework. It introduces a Chameleon hash-based proof of federated unlearning to verifiably erase target clients' data effects without exposing historical model updates, coupled with an adaptive retraining mechanism that estimates each client's contribution to minimize unlearning rounds and computational cost. The framework operates with on-chain smart contracts and off-chain storage, ensuring tamper-evident, auditable unlearning workflows and robust security properties including collision resistance and key-exposure freshness. Experimental results on MNIST, FMNIST, and CIFAR demonstrate improved data removal verification, reduced retraining overhead, and competitive accuracy relative to baselines, signaling practical potential for trustworthy federated unlearning in decentralized settings.

Abstract

Federated unlearning is a promising paradigm for protecting the data ownership of distributed clients. It allows central servers to remove historical data effects within the machine learning model as well as address the "right to be forgotten" issue in federated learning. However, existing works require central servers to retain the historical model parameters from distributed clients, such that allows the central server to utilize these parameters for further training even, after the clients exit the training process. To address this issue, this paper proposes a new blockchain-enabled trustworthy federated unlearning framework. We first design a proof of federated unlearning protocol, which utilizes the Chameleon hash function to verify data removal and eliminate the data contributions stored in other clients' models. Then, an adaptive contribution-based retraining mechanism is developed to reduce the computational overhead and significantly improve the training efficiency. Extensive experiments demonstrate that the proposed framework can achieve a better data removal effect than the state-of-the-art frameworks, marking a significant stride towards trustworthy federated unlearning.

Figures (5)

• Figure 1: Blockchain-enabled trustworthy federated unlearning framework.
• Figure 2: On-chain and off-chain interaction time.
• Figure 3: Accuracy and unlearning performance.
• Figure 4: Time consumption and model deviation.
• Figure 5: Performance on the generation task.