REGLO: Provable Neural Network Repair for Global Robustness Properties

Feisi Fu, Zhilu Wang, Jiameng Fan, Yixuan Wang, Chao Huang, Xin Chen, Qi Zhu, Wenchao Li

Keywords: Neural Network Repair, Global Robustness, Individual Fairness

TL;DR: We present REGLO, the first technique that enables provable repair of DNNs for satisfying global robustness properties.

Abstract: We present REGLO, a novel methodology for repairing neural networks to satisfy global robustness properties. In contrast to existing works that focus on local robustness, i.e., robustness of individual inputs, REGLO tackles global robustness, a strictly stronger notion that requires robustness for all inputs within a region. Leveraging an observation that any counterexample to a global robustness property must exhibit a corresponding large gradient, REGLO first identifies violating regions where the counterexamples reside, then uses verified robustness bounds on these regions to formulate a robust optimization problem to compute a minimal weight change in the network that will provably repair the violations. Experimental results demonstrate the effectiveness of REGLO across a set of benchmarks.