Investigating and Simplifying Masking-based Saliency Methods for Model Interpretability
arXiv:2010.09750
Abstract
Saliency maps that identify the most informative regions of an image for a classifier are valuable for model interpretability. A common approach to creating saliency maps involves generating input masks that mask out portions of an image to maximally deteriorate classification performance, or mask in an image to preserve classification performance. Many variants of this approach have been proposed in the literature, such as counterfactual generation and optimizing over a Gumbel-Softmax distribution. Using a general formulation of masking-based saliency methods, we conduct an extensive evaluation study of a number of recently proposed variants to understand which elements of these methods meaningfully improve performance. Surprisingly, we find that a well-tuned, relatively simple formulation of a masking-based saliency model outperforms many more complex approaches. We find that the most important ingredients for high quality saliency map generation are (1) using both masked-in and masked-out objectives and (2) training the classifier alongside the masking model. Strikingly, we show that a masking model can be trained with as few as 10 examples per class and still generate saliency maps with only a 0.7-point increase in localization error.
References in corpus (13)
- Categorical Reparameterization with Gumbel-Softmax
- Towards A Rigorous Science of Interpretable Machine Learning
- Axiomatic Attribution for Deep Networks
- The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
- Explainable Artificial Intelligence: Understanding, Visualizing and Interpreting Deep Learning Models
- SmoothGrad: removing noise by adding noise
- Sanity Checks for Saliency Maps
- Consistent Individualized Feature Attribution for Tree Ensembles
- Visualizing Deep Neural Network Decisions: Prediction Difference Analysis
- Generative Image Inpainting with Contextual Attention
- RISE: Randomized Input Sampling for Explanation of Black-box Models
- Explaining Image Classifiers by Counterfactual Generation
- Classifier-agnostic saliency map extraction