# SEEV: Synthesis with Efficient Exact Verification for ReLU Neural Barrier Functions

Source: [SEEV: Synthesis with Efficient Exact Verification for ReLU Neural Barrier Functions](<https://www.sota2.com/research/paper/68fa75711e209131cea67f97>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-02T07:37:46.432Z

Authors: Hongchao Zhang, Zhizhen Qin, Sicun Gao, Andrew Clark

Published: 2024-10-27T03:52:03.000Z

## Abstract

Neural Control Barrier Functions (NCBFs) have shown significant promise in enforcing safety constraints on nonlinear autonomous systems. State-of-the-art exact approaches to verifying safety of NCBF-based controllers exploit the piecewise-linear structure of ReLU neural networks, however, such approaches still rely on enumerating all of the activation regions of the network near the safety boundary, thus incurring high computation cost. In this paper, we propose a framework for Synthesis with Efficient Exact Verification (SEEV). Our framework consists of two components, namely (i) an NCBF synthesis algorithm that introduces a novel regularizer to reduce the number of activation regions at the safety boundary, and (ii) a verification algorithm that exploits tight over-approximations of the safety conditions to reduce the cost of verifying each piecewise-linear segment. Our simulations show that SEEV significantly improves verification efficiency while maintaining the CBF quality across various benchmark systems and neural network structures. Our code is available at https://github.com/HongchaoZhang-HZ/SEEV.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [NCBF Verification on OA 1.0 (test)](<https://www.sota2.com/research/sota/ncbf-verification-on-oa-1-0-test.md>): NCBF Verification; OA 1.0 (test); Verification Run-time (s): 0.39.

- [NCBF Verification on Darboux 1.0 (test)](<https://www.sota2.com/research/sota/ncbf-verification-on-darboux-1-0-test.md>): NCBF Verification; Darboux 1.0 (test); Verification Run-time (s): 2.5.

- [NCBF Verification on SR 1.0 (test)](<https://www.sota2.com/research/sota/ncbf-verification-on-sr-1-0-test.md>): NCBF Verification; SR 1.0 (test); Verification Run-time (s): 9.8.

- [NCBF Verification on hi-ord8 1.0 (test)](<https://www.sota2.com/research/sota/ncbf-verification-on-hi-ord8-1-0-test.md>): NCBF Verification; hi-ord8 1.0 (test); Verification Run-time (s): 0.54.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2410.20326>)

[PDF](<https://arxiv.org/pdf/2410.20326.pdf>)

[Implementation](<https://github.com/HongchaoZhang-HZ/SEEV>)
