English

Bridging Dimensions: Confident Reachability for High-Dimensional Controllers

Machine Learning 2024-05-03 v4

Abstract

Autonomous systems are increasingly implemented using end-to-end learning-based controllers. Such controllers make decisions that are executed on the real system, with images as one of the primary sensing modalities. Deep neural networks form a fundamental building block of such controllers. Unfortunately, the existing neural-network verification tools do not scale to inputs with thousands of dimensions -- especially when the individual inputs (such as pixels) are devoid of clear physical meaning. This paper takes a step towards connecting exhaustive closed-loop verification with high-dimensional controllers. Our key insight is that the behavior of a high-dimensional controller can be approximated with several low-dimensional controllers. To balance the approximation accuracy and verifiability of our low-dimensional controllers, we leverage the latest verification-aware knowledge distillation. Then, we inflate low-dimensional reachability results with statistical approximation errors, yielding a high-confidence reachability guarantee for the high-dimensional controller. We investigate two inflation techniques -- based on trajectories and control actions -- both of which show convincing performance in three OpenAI gym benchmarks.

Keywords

Cite

@article{arxiv.2311.04843,
  title  = {Bridging Dimensions: Confident Reachability for High-Dimensional Controllers},
  author = {Yuang Geng and Jake Brandon Baldauf and Souradeep Dutta and Chao Huang and Ivan Ruchkin},
  journal= {arXiv preprint arXiv:2311.04843},
  year   = {2024}
}
R2 v1 2026-06-28T13:15:21.788Z