A New Strategy for Verifying Reach-Avoid Specifications in Neural Feedback Systems

2026-01-12

AI Summary

Existing verification approaches for reach-avoid properties in neural feedback systems predominantly rely on forward reachability analysis, which struggles to balance precision and scalability. This work proposes a unified verification framework that synergistically integrates forward and backward analysis. For the first time, it enables efficient computation of both upper and lower approximations of backward reachable sets and seamlessly combines them with forward methods, thereby overcoming the limitations inherent in traditional unidirectional verification. The proposed approach significantly enhances both completeness and scalability of verification, allowing for more precise treatment of reach-avoid properties in complex neural feedback systems while rigorously guaranteeing safety.

Technology Category

• Multiagent Systems: Adversarial Agents
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Forward reachability analysis is the predominant approach for verifying reach-avoid properties in neural feedback systems (dynamical systems controlled by neural networks). This dominance stems from the limited scalability of existing backward reachability methods. In this work, we introduce new algorithms that compute both over- and under-approximations of backward reachable sets for such systems. We further integrate these backward algorithms with established forward analysis techniques to yield a unified verification framework for neural feedback systems.

Problem

• reach-avoid
• neural feedback systems
• backward reachability
• verification
• scalability

Innovation

• backward reachability
• neural feedback systems
• reach-avoid verification
• over-approximation
• under-approximation