A Literature Review on Verification and Abstraction of Neural Networks Within the Formal Methods Community

Sudeep Kanav, Jan Křetínský, Sabine Rieder

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

Abstract

With the increasing interest in applying neural networks (NNs) to safety-critical problems like autonomous driving or unmanned aircrafts, ensuring the reliability of these NNs becomes essential. Therefore, several verification techniques have been proposed in recent years. Additionally, various abstraction techniques have been developed to enable verification of larger NNs. As the area develops, different surveys on verification of NNs are being published. However, we are missing a systematic summarization of knowledge through the lens of formal methods. In this literature review, we provide a systematic overview of techniques for verification and abstraction of NNs published in well-known formal verification conferences during the last ten years.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Science and Business Media Deutschland GmbH
Pages39-65
Number of pages27
DOIs
StatePublished - 2025

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume15262 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Fingerprint

Dive into the research topics of 'A Literature Review on Verification and Abstraction of Neural Networks Within the Formal Methods Community'. Together they form a unique fingerprint.

Cite this