Breaking Neural Reasoning Architectures With Metamorphic Relation-Based Adversarial Examples

Alvin Chan, Lei Ma, Felix Juefei-Xu, Yew Soon Ong, Xiaofei Xie, Minhui Xue, Yang Liu

研究成果: ジャーナルへの寄稿学術誌査読

4 被引用数 (Scopus)


The ability to read, reason, and infer lies at the heart of neural reasoning architectures. After all, the ability to perform logical reasoning over language remains a coveted goal of Artificial Intelligence. To this end, models such as the Turing-complete differentiable neural computer (DNC) boast of real logical reasoning capabilities, along with the ability to reason beyond simple surface-level matching. In this brief, we propose the first probe into DNC's logical reasoning capabilities with a focus on text-based question answering (QA). More concretely, we propose a conceptually simple but effective adversarial attack based on metamorphic relations. Our proposed adversarial attack reduces DNCs' state-of-the-art accuracy from 100% to 1.5% in the worst case, exposing weaknesses and susceptibilities in modern neural reasoning architectures. We further empirically explore possibilities to defend against such attacks and demonstrate the utility of our adversarial framework as a simple scalable method to improve model adversarial robustness.

ジャーナルIEEE Transactions on Neural Networks and Learning Systems
出版ステータス印刷中 - 2021

!!!All Science Journal Classification (ASJC) codes

  • ソフトウェア
  • コンピュータ サイエンスの応用
  • コンピュータ ネットワークおよび通信
  • 人工知能


「Breaking Neural Reasoning Architectures With Metamorphic Relation-Based Adversarial Examples」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。