Work

Mutual empowerment of distributed AI and blockchain

Year: 2025
Type: article
Abstract: Due to their security, transparency, and decentralized trust model, blockchains have been integrated into a wide range of applications. Distributed and collaborative deep learning approaches, such as Federated Learning and Split Learning, are among the most prominent areas jointly explored with blockchain. This thesis investigates how blockchain mutually empowers distributed learning frameworks—and vice versa—through three novel contributions. First, we propose Proof-of-Collaborative-Learning (PoCL), a blockchain consensus mechanism that replaces energy-intensive mining with productive federated learning tasks, enabling secure, incentive-aligned, and resource-efficient consensus. Second, we advance SplitFed Learning (SFL) by introducing Sharded SFL for improved scalability, and Blockchain-enabled SFL (BSFL), a decentralized variant that uses smart contracts to coordinate shard aggregation and ensure model integrity. Third, we present Blockchain-enabled Personalized Federated Learning (BPFL), a framework that enhances fairness and model ownership by incorporating contribution-based personalization and a tokenized access mechanism, with theoretical convergence guarantees. Through these contributions, this thesis demonstrates how decentralized learning and blockchain can be integrated to improve security, efficiency, and fairness in collaborative AI systems.
Language: English
Open Access status: green
OpenAlex ID: W7137545446
FWCI:
Cites: 70
Cited by:
Related to: 10