Visible to the public Detecting Adversarial Examples for Deep Neural Networks via Layer Directed Discriminative Noise Injection

TitleDetecting Adversarial Examples for Deep Neural Networks via Layer Directed Discriminative Noise Injection
Publication TypeConference Paper
Year of Publication2019
AuthorsWang, Si, Liu, Wenye, Chang, Chip-Hong
Conference Name2019 Asian Hardware Oriented Security and Trust Symposium (AsianHOST)
Date Publisheddec
Keywordsadversarial examples, adversarial images, Computer architecture, Computer vision, computer vision tasks, convolutional neural nets, Deep Learning, deep neural networks, discriminative noise injection strategy, distortion, dominant layers, false positive rate, false trust, layer directed discriminative noise, learning (artificial intelligence), machine learning, MobileNet, natural images, natural scenes, Neural networks, noninvasive universal perturbation attack, Perturbation methods, policy-based governance, Policy-Governed Secure Collaboration, pubcrawl, resilience, Resiliency, Scalability, Sensitivity, Training
Abstract

Deep learning is a popular powerful machine learning solution to the computer vision tasks. The most criticized vulnerability of deep learning is its poor tolerance towards adversarial images obtained by deliberately adding imperceptibly small perturbations to the clean inputs. Such negatives can delude a classifier into wrong decision making. Previous defensive techniques mostly focused on refining the models or input transformation. They are either implemented only with small datasets or shown to have limited success. Furthermore, they are rarely scrutinized from the hardware perspective despite Artificial Intelligence (AI) on a chip is a roadmap for embedded intelligence everywhere. In this paper we propose a new discriminative noise injection strategy to adaptively select a few dominant layers and progressively discriminate adversarial from benign inputs. This is made possible by evaluating the differences in label change rate from both adversarial and natural images by injecting different amount of noise into the weights of individual layers in the model. The approach is evaluated on the ImageNet Dataset with 8-bit truncated models for the state-of-the-art DNN architectures. The results show a high detection rate of up to 88.00% with only approximately 5% of false positive rate for MobileNet. Both detection rate and false positive rate have been improved well above existing advanced defenses against the most practical noninvasive universal perturbation attack on deep learning based AI chip.

URLhttps://ieeexplore.ieee.org/document/9006702
DOI10.1109/AsianHOST47458.2019.9006702
Citation Keywang_detecting_2019