Timezone: »

Quasi-global Momentum: Accelerating Decentralized Deep Learning on Heterogeneous Data
Tao Lin · Sai Praneeth Reddy Karimireddy · Sebastian Stich · Martin Jaggi

Wed Jul 21 09:00 PM -- 11:00 PM (PDT) @ None #None

Decentralized training of deep learning models is a key element for enabling data privacy and on-device learning over networks. In realistic learning scenarios, the presence of heterogeneity across different clients' local datasets poses an optimization challenge and may severely deteriorate the generalization performance. In this paper, we investigate and identify the limitation of several decentralized optimization algorithms for different degrees of data heterogeneity. We propose a novel momentum-based method to mitigate this decentralized training difficulty. We show in extensive empirical experiments on various CV/NLP datasets (CIFAR-10, ImageNet, and AG News) and several network topologies (Ring and Social Network) that our method is much more robust to the heterogeneity of clients' data than other existing methods, by a significant improvement in test performance (1%-20%).

Author Information

Tao Lin (EPFL)
Praneeth Karimireddy (EPFL)
Sebastian Stich (EPFL)
Martin Jaggi (EPFL)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors