Scale-Robust Timely Asynchronous Decentralized Learning
This work addresses scalability challenges in decentralized machine learning for large networks of devices, providing theoretical guarantees for convergence, but it is incremental as it builds on existing asynchronous and gossip-based methods.
The paper tackles the problem of ensuring convergence in asynchronous decentralized learning systems with many devices by establishing a staleness criterion, showing that if each user's gossip capacity scales as Ω(log n), convergence is guaranteed in finite time, while any distributed opportunistic scheme requires Ω(n) scaling to ensure bounded staleness.
We consider an asynchronous decentralized learning system, which consists of a network of connected devices trying to learn a machine learning model without any centralized parameter server. The users in the network have their own local training data, which is used for learning across all the nodes in the network. The learning method consists of two processes, evolving simultaneously without any necessary synchronization. The first process is the model update, where the users update their local model via a fixed number of stochastic gradient descent steps. The second process is model mixing, where the users communicate with each other via randomized gossiping to exchange their models and average them to reach consensus. In this work, we investigate the staleness criteria for such a system, which is a sufficient condition for convergence of individual user models. We show that for network scaling, i.e., when the number of user devices $n$ is very large, if the gossip capacity of individual users scales as $Ω(\log n)$, we can guarantee the convergence of user models in finite time. Furthermore, we show that the bounded staleness can only be guaranteed by any distributed opportunistic scheme by $Ω(n)$ scaling.