×
This paper proposes a method to repair unsafe ReLU DNNs in safety-critical systems using reachability analysis. Our repair method uses reachability analysis to calculate the unsafe reachable domain of a DNN, and then uses a novel loss function to construct its distance to the safe domain during the retraining process.
Aug 29, 2022
Aug 9, 2021 · (4). The framework is evaluated on two benchmark problems where the detailed evolution of model candidates under repair is thoroughly analyzed.
Aug 9, 2021 · This research proposes a framework to repair unsafe DNNs in safety-critical systems with reachability analysis. The repair process is inspired ...
Sep 12, 2022 · This paper proposes a method to repair unsafe ReLU DNNs in safety-critical systems using reachability analysis. Our repair method uses ...
With reaching analysis, Veritex [49] can repair unsafe neural networks that are used in safety-critical systems. It calculates the reachable domain of unsafe ...
Our repair method uses reachability analysis to calculate the unsafe reachable domain of a DNN, and then uses a novel loss function to construct the distance to ...
Jun 14, 2022 · Our repair method uses reachability analysis to calculate the unsafe reachable domain of a DNN, and then uses a novel loss function to construct ...
Johnson, Danil Prokhorov, "Neural Network Repair with Reachability Analysis", In 20th International Conference on Formal Modeling and Analysis of Timed Systems ...
Neural Network Repair with Reachability Analysis ... Authors: Xiaodong Yang; Tom Yamaguchi; Hoang-Dung Tran; Bardh Hoxha; Taylor T. Johnson; Danil Prokhorov. List ...
In this paper, we introduce two new approaches for the reachability analysis of neural networks with additional uncertainties on their internal parameters ( ...