Defending Adversarial Attacks via Semantic Feature Manipulation

Abstract

Machine learning models have demonstrated vulnerability to adversarial attacks, more specifically misclassification of adversarial examples. In this paper, we propose a one-off and attack-agnostic Feature Manipulation (FM)-Defense to detect and purify adversarial examples in an interpretable and efficient manner. The intuition is that the classification result of a normal image is generally resistant to non-significant intrinsic feature changes, e.g., varying thickness of handwritten digits. In contrast, adversarial examples are sensitive to such changes since the perturbation lacks transferability. To enable manipulation of features, a combo-variational autoencoder is applied to learn disentangled latent codes that reveal semantic features. The resistance to classification change over the morphs, derived by varying and reconstructing latent codes, is used to detect suspicious inputs. Further, combo-VAE is enhanced to purify the adversarial examples with good quality by considering both class-shared and class-unique features. We empirically demonstrate the effectiveness of detection and the quality of purified instance. Our experiments on three datasets show that FM-Defense can detect nearly $100\%$ of adversarial examples produced by different state-of-the-art adversarial attacks. It achieves more than $99\%$ overall purification accuracy on the suspicious instances that close the manifold of normal examples.

arXiv commentarXiv admin note: text overlap with arXiv:2001.06640 and text overlap with arXiv:1705.09064 by other authors

References in corpus (10)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• Delving into Transferable Adversarial Examples and Black-box Attacks
• Disentangling by Factorising
• Deep Unsupervised Clustering with Gaussian Mixture Variational Autoencoders
• Understanding Adversarial Training: Increasing Local Stability of Neural Nets through Robust Optimization
• On the (Statistical) Detection of Adversarial Examples
• Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Models
• Featurized Bidirectional GAN: Adversarial Defense via Adversarially Learned Semantic Inference