Pushing the Limit of 1-Minimality of Language-Agnostic Program Reduction
Offered By: ACM SIGPLAN via YouTube
Course Description
Overview
Explore a groundbreaking conference talk on pushing the boundaries of language-agnostic program reduction. Delve into the innovative Vulcan framework, designed to minimize the results of language-agnostic program reducers (AGRs) by leveraging formal language syntax for aggressive program transformations. Discover how this approach creates new reduction opportunities and directly eliminates bug-irrelevant elements from AGR results. Learn about the key insights driving this research, including the limitations of existing program reducer transformations and the potential for diverse transformations in smaller AGR results. Examine the extensive evaluations conducted on multilingual benchmarks, showcasing Vulcan's effectiveness in outperforming state-of-the-art reducers across C, Rust, and SMT-LIBv2 programs. Gain insights into how Vulcan achieves significant token reductions compared to Perses and even improves upon C-Reduce results for C programs. This 17-minute video presentation from OOPSLA 2023 offers valuable knowledge for researchers and practitioners in program reduction, automated debugging, and test input minimization.
Syllabus
[OOPSLA23] Pushing the Limit of 1-Minimality of Language-Agnostic Program Reduction
Taught by
ACM SIGPLAN
Related Courses
Automata TheoryStanford University via edX Computability, Complexity & Algorithms
Georgia Institute of Technology via Udacity Advanced Algorithms and Complexity
University of California, San Diego via Coursera NP-Complete Problems
University of California, San Diego via edX Razonamiento artificial
Universidad Nacional Autónoma de México via Coursera