Minimal Multi-Layer Modifications of Deep Neural Networks

Idan Refaeli, Guy Katz

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Deep neural networks (DNNs) have become increasingly popular in recent years. However, despite their many successes, DNNs may also err and produce incorrect and potentially fatal outputs in safety-critical settings, such as autonomous driving, medical diagnosis, and airborne collision avoidance systems. Much work has been put into detecting such erroneous behavior in DNNs, e.g., via testing or verification, but removing these errors after their detection has received lesser attention. We present here a new tool, called 3M-DNN, for repairing a given DNN, which is known to err on some set of inputs. The novel repair procedure implemented in 3M-DNN computes a modification to the network’s weights that corrects its behavior, and attempts to minimize this change via a sequence of calls to a backend, black-box DNN verification engine. To the best of our knowledge, our method is the first one that allows repairing the network by simultaneously modifying multiple layers. This is achieved by splitting the network into sub-networks, and applying a single-layer repairing technique to each component. We evaluated 3M-DNN tool on an extensive set of benchmarks, obtaining promising results.

Original languageEnglish
Title of host publicationSoftware Verification and Formal Methods for ML-Enabled Autonomous Systems - 5th International Workshop, FoMLAS 2022, and 15th International Workshop, NSV 2022, Proceedings
EditorsOmri Isac, Guy Katz, Radoslav Ivanov, Nina Narodytska, Laura Nenzi
PublisherSpringer Science and Business Media Deutschland GmbH
Pages46-66
Number of pages21
ISBN (Print)9783031212215
DOIs
StatePublished - 2022
Event5th International Workshop on Software Verification and Formal Methods for ML-Enables Autonomous Systems, FoMLAS 2022 and 15th International Workshop on Numerical Software Verification, NSV 2022 - Haifa, Israel
Duration: 11 Aug 202211 Aug 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13466 LNCS

Conference

Conference5th International Workshop on Software Verification and Formal Methods for ML-Enables Autonomous Systems, FoMLAS 2022 and 15th International Workshop on Numerical Software Verification, NSV 2022
Country/TerritoryIsrael
CityHaifa
Period11/08/2211/08/22

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Minimal Multi-Layer Modifications of Deep Neural Networks'. Together they form a unique fingerprint.

Cite this