The 6th International Verification of Neural Networks Competition (VNN-COMP'25), to be held with the 8th International Symposium on AI Verification (SAIV'25) at the 37th International Conference on Computer Aided Verification (CAV'25) over July 21-22, 2025, in Zagreb, Croatia, aims to bring together researchers interested in formal methods and tools providing guarantees about the behaviors of neural networks and systems built from them. Participation will again be conducted remotely in advance of the conference and attendance at the conference will not be required, but results will be finalized and presented there.
The website has been updated with tentative dates for the 2025 edition, which will overall follow similar procedures as the prior couple iterations. We plan one change to encourage participation, in that participants in the competition will have the opportunity to submit and publish with the SAIV Springer LNCS proceedings a "Competition Contribution" paper describing their contributed tool or benchmark, inspired by such papers presented at SV-COMP. SAIV also has a paper track on benchmarks and case studies for more in depth benchmark contributions participants may consider. https://sites.google.com/view/vnn2025 SAIV website: https://www.aiverification.org/2025/ If you are interested to participate with a tool, propose a benchmark, or participate otherwise, please submit the registration/interest form here so we can keep you apprised of next steps. https://forms.gle/73samugGv91woo169 Important Dates: Intention to participate (you may join after this date, but sooner is better to be kept informed): March 24, 2025 Rules and kickoff meeting (Zoom): April 3, 2025, 9am US central time Finalization of the rules: May 1, 2025 Submission of new benchmarks: May 8, 2025 Voting for selecting benchmarks for the competition: June 2, 2025 Participants finalize tool scripts and organizers begin running tools: June 30, 2025 SAIV with presentation of VNN-COMP results and report: July 21-22, 2025 Finally, if you are interested to be more involved in the organization of the competition this iteration or have any suggestions, please let us know by email. Best regards, Taylor (on behalf of the organizers) --- Taylor T. Johnson, PhD, PE Director, VeriVITAL - The Verification and Validation for Intelligent and Trustworthy Autonomy Laboratory (VeriVITAL) Associate Chair, Computer Science (CS) Director of Graduate Studies (DGS), PhD, Computer Science (CS) Director of Master of Engineering (MEng) in Cyber-Physical Systems (CPS) Associate Professor, Computer Science (CS) Associate Professor, Electrical and Computer Engineering (ECE) A. James and Alice B. Clark Foundation Chancellor Faculty Fellow Senior Research Scientist, Institute for Software Integrated Systems Vanderbilt University http://www.TaylorTJohnson.com
