LGCVMANov 17, 2021

Low Precision Decentralized Distributed Training over IID and non-IID Data

arXiv:2111.09389v312 citations
Originality Incremental advance
AI Analysis

This work addresses practical on-device training limitations for edge devices with private data, though it is incremental as it combines existing low precision techniques with decentralized setups.

The paper tackles the communication and compute bottlenecks in decentralized distributed learning by proposing low precision training, which reduces computational complexity, memory usage, and communication cost by 4x and compute energy by ~20x, with less than 1% accuracy loss for both IID and non-IID data.

Decentralized distributed learning is the key to enabling large-scale machine learning (training) on edge devices utilizing private user-generated local data, without relying on the cloud. However, the practical realization of such on-device training is limited by the communication and compute bottleneck. In this paper, we propose and show the convergence of low precision decentralized training that aims to reduce the computational complexity and communication cost of decentralized training. Many feedback-based compression techniques have been proposed in the literature to reduce communication costs. To the best of our knowledge, there is no work that applies and shows compute efficient training techniques such as quantization, pruning, etc., for peer-to-peer decentralized learning setups. Since real-world applications have a significant skew in the data distribution, we design "Range-EvoNorm" as the normalization activation layer which is better suited for low precision training over non-IID data. Moreover, we show that the proposed low precision training can be used in synergy with other communication compression methods decreasing the communication cost further. Our experiments indicate that 8-bit decentralized training has minimal accuracy loss compared to its full precision counterpart even with non-IID data. However, when low precision training is accompanied by communication compression through sparsification we observe a 1-2% drop in accuracy. The proposed low precision decentralized training decreases computational complexity, memory usage, and communication cost by 4x and compute energy by a factor of ~20x, while trading off less than a $1\%$ accuracy for both IID and non-IID data. In particular, with higher skew values, we observe an increase in accuracy (by ~ 0.5%) with low precision training, indicating the regularization effect of the quantization.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes