Deep Competitive Pathway Networks

TL;DR

CoPaNet introduces a novel deep neural network architecture with competitive pathway units that enhance feature learning through parallel subnetworks and pathway encoding, achieving state-of-the-art results on multiple benchmarks.

Contribution

This work proposes the Competitive Pathway Network (CoPaNet), a new architecture that uses parallel residual subnetworks with max operation for improved feature diversity and routing.

Findings

Achieved state-of-the-art results on CIFAR-10 and CIFAR-100.

Performed comparably on SVHN and ImageNet datasets.

Demonstrated the effectiveness of pathway encoding and feature reuse.

Abstract

In the design of deep neural architectures, recent studies have demonstrated the benefits of grouping subnetworks into a larger network. For examples, the Inception architecture integrates multi-scale subnetworks and the residual network can be regarded that a residual unit combines a residual subnetwork with an identity shortcut. In this work, we embrace this observation and propose the Competitive Pathway Network (CoPaNet). The CoPaNet comprises a stack of competitive pathway units and each unit contains multiple parallel residual-type subnetworks followed by a max operation for feature competition. This mechanism enhances the model capability by learning a variety of features in subnetworks. The proposed strategy explicitly shows that the features propagate through pathways in various routing patterns, which is referred to as pathway encoding of category information. Moreover, the cross-block shortcut can be added to the CoPaNet to encourage feature reuse. We evaluated the proposed CoPaNet on four object recognition benchmarks: CIFAR-10, CIFAR-100, SVHN, and ImageNet. CoPaNet obtained the state-of-the-art or comparable results using similar amounts of parameters. The code of CoPaNet is available at: https://github.com/JiaRenChang/CoPaNet.