Home > Research > Publications & Outputs > Reachability analysis of deep neural networks w...
View graph of relations

Reachability analysis of deep neural networks with provable guarantees

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paperpeer-review

Published
Close
Publication date13/07/2018
Host publicationProceedings of the 27th International Joint Conference on Artificial Intelligence, IJCAI 2018
EditorsJerome Lang
PublisherInternational Joint Conferences on Artificial Intelligence
Pages2651-2659
Number of pages9
Volume2018
ISBN (Electronic)9780999241127
<mark>Original language</mark>English
Event27th International Joint Conference on Artificial Intelligence, IJCAI 2018 - Stockholm, Sweden
Duration: 13/07/201819/07/2018

Conference

Conference27th International Joint Conference on Artificial Intelligence, IJCAI 2018
Country/TerritorySweden
CityStockholm
Period13/07/1819/07/18

Conference

Conference27th International Joint Conference on Artificial Intelligence, IJCAI 2018
Country/TerritorySweden
CityStockholm
Period13/07/1819/07/18

Abstract

Verifying correctness of deep neural networks (DNNs) is challenging. We study a generic reachability problem for feed-forward DNNs which, for a given set of inputs to the network and a Lipschitz-continuous function over its outputs, computes the lower and upper bound on the function values. Because the network and the function are Lipschitz continuous, all values in the interval between the lower and upper bound are reachable. We show how to obtain the safety verification problem, the output range analysis problem and a robustness measure by instantiating the reachability problem. We present a novel algorithm based on adaptive nested optimisation to solve the reachability problem. The technique has been implemented and evaluated on a range of DNNs, demonstrating its efficiency, scalability and ability to handle a broader class of networks than state-of-the-art verification approaches.