Real Time Image Saliency for Black Box Classifiers

Abstract

In this work we develop a fast saliency detection method that can be applied to any differentiable image classifier. We train a masking model to manipulate the scores of the classifier by masking salient parts of the input image. Our model generalises well to unseen images and requires a single forward pass to perform saliency detection, therefore suitable for use in real-time systems. We test our approach on CIFAR-10 and ImageNet datasets and show that the produced saliency maps are easily interpretable, sharp, and free of artifacts. We suggest a new metric for saliency and test our method on the ImageNet object localisation task. We achieve results outperforming other weakly supervised methods.

References in corpus (3)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Intriguing properties of neural networks
• Striving for Simplicity: The All Convolutional Net

Cited by in corpus (13)

• Interpretable machine learning: definitions, methods, and applications
• Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)
• Hierarchical interpretations for neural network predictions
• Visualizing Deep Networks by Optimizing with Integrated Gradients
• Deep Weakly-Supervised Learning Methods for Classification and Localization in Histology Images: A Survey
• Interpretations are useful: penalizing explanations to align neural networks with prior knowledge
• Dropout Feature Ranking for Deep Learning Models
• Deeply Explain CNN via Hierarchical Decomposition
• Transformation Importance with Applications to Cosmology
• Can We Faithfully Represent Masked States to Compute Shapley Values on a DNN?
• Interpreting search result rankings through intent modeling
• A Hypothesis for the Aesthetic Appreciation in Neural Networks
• Explainability in subgraphs-enhanced Graph Neural Networks