veriFIRE: an Industrial Case Study in Verifying Consistency Properties for a DNN-Based Wildfire Detection System

2026-06-02

AI Summary

This work addresses the reliability challenges of airborne wildfire detection systems in safety-critical scenarios by proposing an end-to-end approach that translates domain-specific consistency requirements—such as monotonicity of detection confidence with increasing target intensity and bounded response under sensor blur—into formal queries amenable to neural network verification. It represents the first effort to formally specify and verify such semantic properties in an industrial-scale system, effectively bridging the gap between practical safety demands and formal neural network verification. Leveraging existing verification tools, the method achieves efficient large-scale validation on real-world background data: all monotonicity queries are solved within five minutes, and robustness verification under blur demonstrates the feasibility of providing meaningful, domain-relevant guarantees even in high-dimensional input spaces.

Abstract

We present our ongoing work on the veriFIRE project: a collaboration between industry and academia, aimed at applying verification to increase the reliability of a real-world, safety-critical system. Specifically, we target an airborne platform for wildfire detection, which incorporates two deep neural networks. We present an end-to-end methodology for verifying \textit{consistency properties} in this system. Our approach encodes application-grounded requirements into solver-compatible queries for existing neural network verifiers. We study properties of interest over critical operational scenarios: (i) monotonicity of detector confidence as target intensity increases; and (ii) bounded detector response under physically plausible blur over the sensor. We instantiate these encodings using state-of-the-art neural network verification backends and evaluate them at scale on real background samples. For the first property, all verification queries are solved in under five minutes. For the second property, verification is substantially harder, highlighting key scalability challenges for richer, higher-dimensional specifications. Overall, the results demonstrate that meaningful, domain-specific guarantees can be obtained for industrial systems.

Problem

• wildfire detection
• deep neural networks
• consistency properties
• verification
• safety-critical systems

Innovation

• neural network verification
• consistency properties
• wildfire detection
• formal methods
• safety-critical systems