Probabilistic Verification of Neural Networks via Efficient Probabilistic Hull Generation
Jingyang Li, Xin Chen, Hongfei Fu, Guoqiang Li
TLDR
This paper introduces a novel framework for probabilistic verification of neural networks, efficiently computing a guaranteed range for safe probabilities.
Key contributions
- Uses regression trees for state space subdivision to generate probabilistic hulls.
- Employs boundary-aware sampling to precisely identify safety boundaries in the input space.
- Applies iterative refinement with probabilistic prioritization for accurate safe probability ranges.
- Outperforms state-of-the-art methods on benchmarks like ACAS Xu and rocket lander.
Why it matters
Probabilistic verification is crucial for NNs operating with uncertain inputs, like in safety-critical systems. This method offers a more accurate and efficient way to quantify safety probabilities. It improves reliability and trustworthiness of AI applications in real-world scenarios.
Original Abstract
The problem of probabilistic verification of a neural network investigates the probability of satisfying the safe constraints in the output space when the input is given by a probability distribution. It is significant to answer this problem when the input is affected by disturbances often modeled by probabilistic variables. In the paper, we propose a novel neural network probabilistic verification framework which computes a guaranteed range for the safe probability by efficiently finding safe and unsafe probabilistic hulls. Our approach consists of three main innovations: (1) a state space subdivision strategy using regression trees to produce probabilistic hulls, (2) a boundary-aware sampling method which identifies the safety boundary in the input space using samples that are later used for building regression trees, and (3) iterative refinement with probabilistic prioritization for computing a guaranteed range for the safe probability. The accuracy and efficiency of our approach are evaluated on various benchmarks including ACAS Xu and a rocket lander controller. The result shows an obvious advantage over the state of the art.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.