Deep Pyramidal Residual Networks

TL;DR

This paper introduces Deep Pyramidal Residual Networks that gradually increase feature map dimensions across all units, enhancing generalization and accuracy in image classification tasks, outperforming traditional residual networks.

Contribution

The paper proposes a novel residual network architecture with gradually increasing feature map dimensions and a new residual unit, improving classification accuracy and generalization.

Findings

Superior performance on CIFAR-10 and CIFAR-100 datasets.

Outperforms original residual networks on ImageNet.

Code availability for reproducibility.

Abstract

Deep convolutional neural networks (DCNNs) have shown remarkable performance in image classification tasks in recent years. Generally, deep neural network architectures are stacks consisting of a large number of convolutional layers, and they perform downsampling along the spatial dimension via pooling to reduce memory usage. Concurrently, the feature map dimension (i.e., the number of channels) is sharply increased at downsampling locations, which is essential to ensure effective performance because it increases the diversity of high-level attributes. This also applies to residual networks and is very closely related to their performance. In this research, instead of sharply increasing the feature map dimension at units that perform downsampling, we gradually increase the feature map dimension at all units to involve as many locations as possible. This design, which is discussed in depth together with our new insights, has proven to be an effective means of improving generalization ability. Furthermore, we propose a novel residual unit capable of further improving the classification accuracy with our new network architecture. Experiments on benchmark CIFAR-10, CIFAR-100, and ImageNet datasets have shown that our network architecture has superior generalization ability compared to the original residual networks. Code is available at https://github.com/jhkim89/PyramidNet}