A team of researchers from Carnegie Mellon University, Northeastern University, Columbia University, and UCLA won the 2021 International Verification of Neural Networks Competition with an open-source tool that can provide a guarantee of the behavior of a critical part of modern artificial intelligence.
Called α,β-CROWN (alpha-beta-CROWN), the tool is a collection of algorithms that run in parallel. It is general and scalable and performed well on a variety of distinct tasks with diverse neural network structures.
"Neural networks are often black-boxes, and it is hard to guarantee that they behave safely and predictably under noisy or malicious inputs," says team leader Huan Zhang, a postdoctoral researcher at Carnegie Mellon. "α,β-CROWN works at speeds two to three orders of magnitude faster than tools developed just three years ago. This has led to significant progress in neural network verification and could lower barriers to adoption in safety-critical fields."
From Carnegie Mellon University
View Full Article