# Ensemble-Compression: A New Method for Parallel Training of Deep Neural Networks

arXiv: https://arxiv.org/abs/1606.00575 · HTML: https://mlsanity.com/paper/1606.00575/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Jun 2016.

## Built on by (most cited first, 4 of 4)

- [Deep Class Incremental Learning from Decentralized Data](https://mlsanity.com/paper/2203.05984/) (2022)
- [Model Compression for Dynamic Forecast Combination](https://mlsanity.com/paper/2104.01830/) (2021)
- [BWCP: Probabilistic Learning-to-Prune Channels for ConvNets via Batch Whitening](https://mlsanity.com/paper/2105.06423/) (2021)
- [Adaptive Consensus Gradients Aggregation for Scaled Distributed Training](https://mlsanity.com/paper/2411.03742/) (2024)

## Abstract (arXiv)

Parallelization framework has become a necessity to speed up the training of deep neural networks (DNN) recently. Such framework typically employs the Model Average approach, denoted as MA-DNN, in which parallel workers conduct respective training based on their own local data while the parameters of local models are periodically communicated and averaged to obtain a global model which serves as the new start of local models. However, since DNN is a highly non-convex model, averaging parameters cannot ensure that such global model can perform better than those local models. To tackle this problem, we introduce a new parallel training framework called Ensemble-Compression, denoted as EC-DNN. In this framework, we propose to aggregate the local models by ensemble, i.e., averaging the outputs of local models instead of the parameters. As most of prevalent loss functions are convex to the output of DNN, the performance of ensemble-based global model is guaranteed to be at least as good as the average performance of local models. However, a big challenge lies in the explosion of model size since each round of ensemble can give rise to multiple times size increment. Thus, we carry out model compression after each ensemble, specialized by a distillation based method in this paper, to reduce the size of the global model to be the same as the local ones. Our experimental results demonstrate the prominent advantage of EC-DNN over MA-DNN in terms of both accuracy and speedup.