Departmental Papers (CIS)

Date of this Version

4-2019

Document Type

Conference Paper

Abstract

This paper presents Verisig, a hybrid system approach to verifying safety properties of closed-loop systems using neural networks as controllers. We focus on sigmoid-based networks and exploit the fact that the sigmoid is the solution to a quadratic differential equation, which allows us to transform the neural network into an equivalent hybrid system. By composing the network’s hybrid system with the plant’s, we transform the problem into a hybrid system verification problem which can be solved using state-of-theart reachability tools. We show that reachability is decidable for networks with one hidden layer and decidable for general networks if Schanuel’s conjecture is true. We evaluate the applicability and scalability of Verisig in two case studies, one from reinforcement learning and one in which the neural network is used to approximate a model predictive controller.

Subject Area

CPS Safe Autonomy, CPS Formal Methods

Publication Source

22nd ACM International Conference on Hybrid Systems: Computation and Control (HSCC 2019)

Start Page

169

Last Page

178

DOI

10.1145/3302504.3311806

Keywords

Neural Network Verification, Hybrid Systems with Neural Network Controllers, Learning-Enabled Components

Share

COinS
 

Date Posted: 07 March 2020

This document has been peer reviewed.