Using Deep Reinforcement Learning and Formal Verification in Safety Critical Systems: Strategies and Challenges

Satyam Sharma, Muhammad Abdul Basit Ur Rahim, Shahid Hussain, Muhammad Rizwan Abid, Tairan Liu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Deep Reinforcement Learning (DRL) is critical in modern Artificial Intelligence (AI), powering innovations from gaming to autonomous vehicles. As DRL continues its rapid ascent, ensuring its systems are both trustworthy and effective is crucial. This research focuses on different DRL techniques and the challenges faced in real-life scenarios. The paper also describes various formal verification techniques and the challenges related to their application. It sheds light on the different frameworks and tools that can enhance the credibility of systems. We performed an extensive literature survey to present the existing methodologies, tools, and frameworks. The analysis systematically reviews and categorizes various formal verification techniques and frameworks employed in DRL. The insights garnered from this study are anticipated to foster an enriched understanding of the processes and contribute to decision-making in Safety Critical Systems using DRL and verification.

Original languageEnglish (US)
Title of host publicationProceedings - 2023 IEEE 23rd International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages834-842
Number of pages9
ISBN (Electronic)9798350359398
DOIs
StatePublished - 2023
Event23rd IEEE International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023 - Chiang Mai, Thailand
Duration: Oct 22 2023Oct 26 2023

Publication series

NameProceedings - 2023 IEEE 23rd International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023

Conference

Conference23rd IEEE International Conference on Software Quality, Reliability, and Security Companion, QRS-C 2023
Country/TerritoryThailand
CityChiang Mai
Period10/22/2310/26/23

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition
  • Software
  • Safety, Risk, Reliability and Quality

Fingerprint

Dive into the research topics of 'Using Deep Reinforcement Learning and Formal Verification in Safety Critical Systems: Strategies and Challenges'. Together they form a unique fingerprint.

Cite this