An Architecture for Resilient Federated Learning through Parallel Recognition
Published in International Conference on Parallel Architectures and Compilation Techniques (PACT), 2022
In federated learning, non-independent and identically distributed (non-IID) local datasets lead to accuracy loss compared to homogeneous distribution of datasets. In this paper, we propose an architecture for improving accuracy and offering resilience through federation utilizing non-IID datasets. The proposed architecture performs parallel recognition employing triplication of AI processors with different intelligence. Experimental results demonstrate that the proposed architecture improves accuracy by 2.3% compared to accuracy of a single AI processor on average and guarantees resilience.
Jeongeun Kim, Youngwoo Jeong, Suyeon Jang, Seung Eun Lee. "An Architecture for Resilient Federated Learning through Parallel Recognition." International Conference on Parallel Architectures and Compilation Techniques (PACT), 2022.
