Visualizing and Understanding Convolutional Networks

TL;DR

This paper introduces visualization techniques and ablation studies for convolutional networks, providing insights into their functioning and leading to improved architectures that outperform previous models on ImageNet and other datasets.

Contribution

It presents a novel visualization method for understanding intermediate features and performs ablation studies to optimize model architecture for better performance.

Findings

New visualization technique reveals network internals

Optimized architectures outperform previous models on ImageNet

Model generalizes well to other datasets like Caltech-101 and Caltech-256

Abstract

Large Convolutional Network models have recently demonstrated impressive classification performance on the ImageNet benchmark. However there is no clear understanding of why they perform so well, or how they might be improved. In this paper we address both issues. We introduce a novel visualization technique that gives insight into the function of intermediate feature layers and the operation of the classifier. We also perform an ablation study to discover the performance contribution from different model layers. This enables us to find model architectures that outperform Krizhevsky \etal on the ImageNet classification benchmark. We show our ImageNet model generalizes well to other datasets: when the softmax classifier is retrained, it convincingly beats the current state-of-the-art results on Caltech-101 and Caltech-256 datasets.