A Singular Value Perspective on Model Robustness
arXiv:2012.03516
Abstract
Convolutional Neural Networks (CNNs) have made significant progress on several computer vision benchmarks, but are fraught with numerous non-human biases such as vulnerability to adversarial samples. Their lack of explainability makes identification and rectification of these biases difficult, and understanding their generalization behavior remains an open problem. In this work we explore the relationship between the generalization behavior of CNNs and the Singular Value Decomposition (SVD) of images. We show that naturally trained and adversarially robust CNNs exploit highly different features for the same dataset. We demonstrate that these features can be disentangled by SVD for ImageNet and CIFAR-10 trained networks. Finally, we propose Rank Integrated Gradients (RIG), the first rank-based feature attribution method to understand the dependence of CNNs on image rank.
References in corpus (20)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Explaining and Harnessing Adversarial Examples
- Intriguing properties of neural networks
- Deep Residual Learning for Image Recognition
- Striving for Simplicity: The All Convolutional Net
- Densely Connected Convolutional Networks
- Towards Deep Learning Models Resistant to Adversarial Attacks
- Theoretically Principled Trade-off between Robustness and Accuracy
- Linformer: Self-Attention with Linear Complexity
- Certified Defenses against Adversarial Examples
- Robustness May Be at Odds with Accuracy
- Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methods
- Measuring the tendency of CNNs to Learn Surface Statistical Regularities
- Natural Adversarial Examples
- Rethinking Attention with Performers
- DeepFool: a simple and accurate method to fool deep neural networks
- Universal adversarial perturbations
- High Frequency Component Helps Explain the Generalization of Convolutional Neural Networks
- Towards Frequency-Based Explanation for Robust CNN
- Scratch that! An Evolution-based Adversarial Attack against Neural Networks