Publication Type

Journal Article

Version

acceptedVersion

Publication Date

8-2022

Abstract

Deep Neural Networks (DNNs) have been widely adopted, yet DNN models are surprisingly unreliable, which raises significant concerns about their use in critical domains. In this work, we propose that runtime DNN mistakes can be quickly detected and properly dealt with in deployment, especially in settings like self-driving vehicles. Just as software engineering (SE) community has developed effective mechanisms and techniques to monitor and check programmed components, our previous work, SelfChecker, is designed to monitor and correct DNN predictions given unintended abnormal test data. SelfChecker triggers an alarm if the decisions given by the internal layer features of the model are inconsistent with the final prediction and provides advice in the form of an alternative prediction. In this paper, we extend SelfChecker to the security domain. Specifically, we describe SelfChecker++, which we designed to target both unintended abnormal test data and intended adversarial samples. Technically, we develop a technique which can transform any runtime inputs triggering alarms into semantically equivalent inputs, then we feed those transformed inputs to the model. Such runtime transformation nullifies any intended crafted samples, making the model immune to adversarial attacks that craft adversarial samples. We evaluated the alarm accuracy of SelfChecker++ on three DNN models and four popular image datasets, and found that SelfChecker++ triggers correct alarms on 63.10% of wrong DNN predictions, and triggers false alarms on 5.77% of correct DNN predictions. We also evaluated the effectiveness of SelfChecker++ in detecting adversarial examples and found it detects on average 70.09% of such samples with advice accuracy that is 20.89% higher than the original DNN model and 18.37% higher than SelfChecker.

Keywords

self-checking system, trustworthiness, deep neural networks, adversarial examples, deployment

Discipline

Artificial Intelligence and Robotics | OS and Networks

Research Areas

Intelligent Systems and Optimization

Publication

IEEE Transactions on Dependable and Secure Computing

First Page

1

Last Page

17

ISSN

1545-5971

Identifier

10.1109/TDSC.2022.3200421

Publisher

Institute of Electrical and Electronics Engineers

Additional URL

https://doi.org/10.1109/TDSC.2022.3200421

Share

COinS