Recognizing Object by Components with Human Prior Knowledge Enhances Adversarial Robustness of Deep Neural Networks

TL;DR

This paper introduces ROCK, a part-based object recognition model inspired by human cognition, which enhances adversarial robustness of deep neural networks by incorporating human prior knowledge and object part segmentation.

Contribution

The paper proposes a novel part-based recognition model, ROCK, that integrates human prior knowledge and object segmentation to improve adversarial robustness of DNNs.

Findings

ROCK outperforms classical models under attack.

Part-based models enhance adversarial robustness.

Incorporating human prior knowledge is effective.

Abstract

Adversarial attacks can easily fool object recognition systems based on deep neural networks (DNNs). Although many defense methods have been proposed in recent years, most of them can still be adaptively evaded. One reason for the weak adversarial robustness may be that DNNs are only supervised by category labels and do not have part-based inductive bias like the recognition process of humans. Inspired by a well-known theory in cognitive psychology -- recognition-by-components, we propose a novel object recognition model ROCK (Recognizing Object by Components with human prior Knowledge). It first segments parts of objects from images, then scores part segmentation results with predefined human prior knowledge, and finally outputs prediction based on the scores. The first stage of ROCK corresponds to the process of decomposing objects into parts in human vision. The second stage corresponds to the decision process of the human brain. ROCK shows better robustness than classical recognition models across various attack settings. These results encourage researchers to rethink the rationality of currently widely-used DNN-based object recognition models and explore the potential of part-based models, once important but recently ignored, for improving robustness.