A New Method to Visualize Deep Neural Networks

Abstract

We present a method for visualising the response of a deep neural network to a specific input. For image data for instance our method will highlight areas that provide evidence in favor of, and against choosing a certain class. The method overcomes several shortcomings of previous methods and provides great additional insight into the decision making process of convolutional networks, which is important both to improve models and to accelerate the adoption of such methods in e.g. medicine. In experiments on ImageNet data, we illustrate how the method works and can be applied in different ways to understand deep neural nets.

Please note that this version of the article is outdated. The new version (published at ICLR2017) includes additional experiments on MRI scans and can be found at arXiv:1702.04595

References in corpus (4)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Caffe: Convolutional Architecture for Fast Feature Embedding
• Understanding Neural Networks Through Deep Visualization
• Object Detectors Emerge in Deep Scene CNNs

Cited by in corpus (8)

• Smooth Grad-CAM++: An Enhanced Inference Level Visualization Technique for Deep Convolutional Neural Network Models
• Understanding and Comparing Deep Neural Networks for Age and Gender Classification
• Understanding Integrated Gradients with SmoothTaylor for Deep Neural Network Attribution
• Visual pathways from the perspective of cost functions and multi-task deep neural networks
• Opening the black box of neural nets: case studies in stop/top discrimination
• Detecting Adversarial Perturbations with Saliency
• A Step Towards Exposing Bias in Trained Deep Convolutional Neural Network Models
• Understanding Convolutional Neural Networks with A Mathematical Model