Author:
Akintunde Michael E.,Kevorchian Andreea,Lomuscio Alessio,Pirovano Edoardo
Abstract
We introduce agent-environment systems where the agent is stateful and executing a ReLU recurrent neural network. We define and study their verification problem by providing equivalences of recurrent and feed-forward neural networks on bounded execution traces. We give a sound and complete procedure for their verification against properties specified in a simplified version of LTL on bounded executions. We present an implementation and discuss the experimental results obtained.
Publisher
Association for the Advancement of Artificial Intelligence (AAAI)
Cited by
26 articles.
订阅此论文施引文献
订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献
1. Engineering Responsible And Explainable Models In Human-Agent Collectives;Applied Artificial Intelligence;2023-12-05
2. Formal Verification of Long Short-Term Memory based Audio Classifiers: A Star based Approach;Electronic Proceedings in Theoretical Computer Science;2023-11-15
3. Input Validation for Neural Networks via Local Robustness Verification;2023 IEEE 23rd International Conference on Software Quality, Reliability, and Security Companion (QRS-C);2023-10-22
4. A review of abstraction methods towards verifying neural networks;ACM Transactions on Embedded Computing Systems;2023-08-28
5. Verification of Recurrent Neural Networks with Star Reachability;Proceedings of the 26th ACM International Conference on Hybrid Systems: Computation and Control;2023-05-09