Decentralized adaptive clustering of deep nets is beneficial for client collaboration

TL;DR

This paper introduces a decentralized adaptive clustering algorithm for personalized deep learning models that dynamically adapts to client similarities without requiring predefined cluster numbers, improving collaboration in non-i.i.d. data settings.

Contribution

The proposed method uses a novel adaptive gossip algorithm for soft cluster assignment, enabling effective client collaboration without hyperparameter tuning for cluster count.

Findings

Outperforms state-of-the-art algorithms in non-i.i.d. data scenarios

Effectively handles covariate and label shift among clients

Adapts continuously to network topology without fixed cluster parameters

Abstract

We study the problem of training personalized deep learning models in a decentralized peer-to-peer setting, focusing on the setting where data distributions differ between the clients and where different clients have different local learning tasks. We study both covariate and label shift, and our contribution is an algorithm which for each client finds beneficial collaborations based on a similarity estimate for the local task. Our method does not rely on hyperparameters which are hard to estimate, such as the number of client clusters, but rather continuously adapts to the network topology using soft cluster assignment based on a novel adaptive gossip algorithm. We test the proposed method in various settings where data is not independent and identically distributed among the clients. The experimental evaluation shows that the proposed method performs better than previous state-of-the-art algorithms for this problem setting, and handles situations well where previous methods fail.