Multi-Residual Networks: Improving the Speed and Accuracy of Residual Networks

TL;DR

This paper introduces multi-residual networks that enhance residual network architectures by increasing residual functions, leading to wider models that improve accuracy and computational efficiency on image classification tasks.

Contribution

The paper proposes a novel multi-residual network architecture that explicitly exploits ensemble behavior, resulting in wider models with better accuracy and efficiency.

Findings

Achieved 3.73% error on CIFAR-10

Outperformed deep residual networks on ImageNet by 0.22%

Implemented model parallelism with 15% computational gains

Abstract

In this article, we take one step toward understanding the learning behavior of deep residual networks, and supporting the observation that deep residual networks behave like ensembles. We propose a new convolutional neural network architecture which builds upon the success of residual networks by explicitly exploiting the interpretation of very deep networks as an ensemble. The proposed multi-residual network increases the number of residual functions in the residual blocks. Our architecture generates models that are wider, rather than deeper, which significantly improves accuracy. We show that our model achieves an error rate of 3.73% and 19.45% on CIFAR-10 and CIFAR-100 respectively, that outperforms almost all of the existing models. We also demonstrate that our model outperforms very deep residual networks by 0.22% (top-1 error) on the full ImageNet 2012 classification dataset. Additionally, inspired by the parallel structure of multi-residual networks, a model parallelism technique has been investigated. The model parallelism method distributes the computation of residual blocks among the processors, yielding up to 15% computational complexity improvement.