Beyond Finite Layer Neural Networks: Bridging Deep Architectures and Numerical Differential Equations

TL;DR

This paper connects deep neural network architectures with numerical differential equations, proposing a new multi-step design inspired by numerical methods that improves efficiency and accuracy in image classification tasks.

Contribution

It introduces the LM-architecture inspired by linear multi-step methods, enabling more efficient and accurate deep networks with fewer parameters, and links stochastic control to training strategies.

Findings

LM-ResNet/LM-ResNeXt outperform ResNet/ResNeXt in accuracy.

Networks can be compressed by over 50% while maintaining performance.

Stochastic depth enhances generalization of LM-ResNet.

Abstract

In our work, we bridge deep neural network design with numerical differential equations. We show that many effective networks, such as ResNet, PolyNet, FractalNet and RevNet, can be interpreted as different numerical discretizations of differential equations. This finding brings us a brand new perspective on the design of effective deep architectures. We can take advantage of the rich knowledge in numerical analysis to guide us in designing new and potentially more effective deep networks. As an example, we propose a linear multi-step architecture (LM-architecture) which is inspired by the linear multi-step method solving ordinary differential equations. The LM-architecture is an effective structure that can be used on any ResNet-like networks. In particular, we demonstrate that LM-ResNet and LM-ResNeXt (i.e. the networks obtained by applying the LM-architecture on ResNet and ResNeXt respectively) can achieve noticeably higher accuracy than ResNet and ResNeXt on both CIFAR and ImageNet with comparable numbers of trainable parameters. In particular, on both CIFAR and ImageNet, LM-ResNet/LM-ResNeXt can significantly compress ($>50$\%) the original networks while maintaining a similar performance. This can be explained mathematically using the concept of modified equation from numerical analysis. Last but not least, we also establish a connection between stochastic control and noise injection in the training process which helps to improve generalization of the networks. Furthermore, by relating stochastic training strategy with stochastic dynamic system, we can easily apply stochastic training to the networks with the LM-architecture. As an example, we introduced stochastic depth to LM-ResNet and achieve significant improvement over the original LM-ResNet on CIFAR10.