PolyNet: A Pursuit of Structural Diversity in Very Deep Networks

Abstract

A number of studies have shown that increasing the depth or width of convolutional networks is a rewarding approach to improve the performance of image recognition. In our study, however, we observed difficulties along both directions. On one hand, the pursuit for very deep networks is met with a diminishing return and increased training difficulty; on the other hand, widening a network would result in a quadratic growth in both computational cost and memory demand. These difficulties motivate us to explore structural diversity in designing deep networks, a new dimension beyond just depth and width. Specifically, we present a new family of modules, namely the PolyInception, which can be flexibly inserted in isolation or in a composition as replacements of different parts of a network. Choosing PolyInception modules with the guidance of architectural efficiency can improve the expressive power while preserving comparable computational cost. The Very Deep PolyNet, designed following this direction, demonstrates substantial improvements over the state-of-the-art on the ILSVRC 2012 benchmark. Compared to Inception-ResNet-v2, it reduces the top-5 validation error on single crops from 4.9% to 4.25%, and that on multi-crops from 3.7% to 3.45%.

Tech report

References in corpus (11)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Densely Connected Convolutional Networks
• Wide Residual Networks
• Resnet in Resnet: Generalizing Residual Architectures
• FractalNet: Ultra-Deep Neural Networks without Residuals
• Residual Networks Behave Like Ensembles of Relatively Shallow Networks
• Residual Networks of Residual Networks: Multilevel Residual Networks
• Systematic evaluation of CNN advances on the ImageNet
• CUHK & ETHZ & SIAT Submission to ActivityNet Challenge 2016
• Swapout: Learning an ensemble of deep architectures
• Multi-Residual Networks: Improving the Speed and Accuracy of Residual Networks

Cited by in corpus (13)

• The History Began from AlexNet: A Comprehensive Survey on Deep Learning Approaches
• Learning Spatio-Temporal Representation with Pseudo-3D Residual Networks
• MgNet: A Unified Framework of Multigrid and Convolutional Neural Network
• Multi-Residual Networks: Improving the Speed and Accuracy of Residual Networks
• Comparative evaluation of CNN architectures for Image Caption Generation
• Sharing Residual Units Through Collective Tensor Factorization in Deep Neural Networks
• Classification of Skin Disease Using Transfer Learning in Convolutional Neural Networks
• Training Group Orthogonal Neural Networks with Privileged Information
• Between-class Learning for Image Classification
• AOGNets: Compositional Grammatical Architectures for Deep Learning
• Multilevel Context Representation for Improving Object Recognition
• Tandem Blocks in Deep Convolutional Neural Networks
• Deep Competitive Pathway Networks