Kartikeya Bhardwaj

FedMAX: Mitigating Activation Divergence for Accurate and Communication-Efficient Federated Learning

2020 · arXiv (Cornell University)

In this paper, we identify a new phenomenon called activation-divergence which occurs in Federated Learning (FL) due to data heterogeneity (i.e., data being non-IID) across multiple users. Specifically, we argue that the activation vectors in …