Visualizing and Understanding Convolutional Networks
This work addresses the interpretability and optimization of convolutional networks for computer vision researchers, providing insights and improved performance on key benchmarks.
The paper tackled the lack of understanding of why large convolutional networks perform well on ImageNet and how to improve them, by introducing a visualization technique and ablation study that led to model architectures outperforming Krizhevsky et al. on ImageNet and achieving state-of-the-art results on Caltech-101 and Caltech-256 datasets.
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.