Is Robustness the Cost of Accuracy? -- A Comprehensive Study on the Robustness of 18 Deep Image Classification Models

Abstract

The prediction accuracy has been the long-lasting and sole standard for comparing the performance of different image classification models, including the ImageNet competition. However, recent studies have highlighted the lack of robustness in well-trained deep neural networks to adversarial examples. Visually imperceptible perturbations to natural images can easily be crafted and mislead the image classifiers towards misclassification. To demystify the trade-offs between robustness and accuracy, in this paper we thoroughly benchmark 18 ImageNet models using multiple robustness metrics, including the distortion, success rate and transferability of adversarial examples between 306 pairs of models. Our extensive experimental results reveal several new insights: (1) linear scaling law - the empirical ${\ell }_2$ and ${\ell }_{\infty }$ distortion metrics scale linearly with the logarithm of classification error; (2) model architecture is a more critical factor to robustness than model size, and the disclosed accuracy-robustness Pareto frontier can be used as an evaluation criterion for ImageNet model designers; (3) for a similar network architecture, increasing network depth slightly improves robustness in ${\ell }_{\infty }$ distortion; (4) there exist models (in VGG family) that exhibit high adversarial transferability, while most adversarial examples crafted from one model can only be transferred within the same family. Experiment code is publicly available at \url{https://github.com/huanzhang12/Adversarial_Survey}.

arXiv commentAccepted by the European Conference on Computer Vision (ECCV) 2018

References in corpus (12)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Neural Architecture Search with Reinforcement Learning
• ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Models
• Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samples
• Delving into Transferable Adversarial Examples and Black-box Attacks
• Formal Guarantees on the Robustness of a Classifier against Adversarial Manipulation
• Spatially Transformed Adversarial Examples
• Adversarial examples from computational constraints
• Query-Efficient Hard-label Black-box Attack:An Optimization-based Approach
• Attacking the Madry Defense Model with $L_1$-based Adversarial Examples
• On the Limitation of Local Intrinsic Dimensionality for Characterizing the Subspaces of Adversarial Examples

Cited by in corpus (4)

• Characterizing Audio Adversarial Examples Using Temporal Dependency
• Defensive Quantization: When Efficiency Meets Robustness
• Is Ordered Weighted ${\ell }_1$ Regularized Regression Robust to Adversarial Perturbation? A Case Study on OSCAR
• On The Utility of Conditional Generation Based Mutual Information for Characterizing Adversarial Subspaces