Design Specification
Design specification research focuses on accurately and efficiently defining the desired characteristics and behavior of systems, encompassing diverse applications from robotic systems to software and hardware. Current efforts concentrate on improving the robustness and generalization of specifications, particularly addressing issues like specification gaming in AI and developing methods for efficient specification learning from limited data, often employing techniques like Bayesian optimization and gradient boosted regression trees. These advancements are crucial for improving the reliability and performance of complex systems across various domains, ranging from manufacturing and construction to AI safety and healthcare.
Papers
Evaluating the Ability of Large Language Models to Generate Verifiable Specifications in VeriFast
Marilyn Rego, Wen Fan, Xin Hu, Sanya Dod, Zhaorui Ni, Danning Xie, Jenna DiVincenzo, Lin Tan
Show, Don't Tell: Learning Reward Machines from Demonstrations for Reinforcement Learning-Based Cardiac Pacemaker Synthesis
John Komp, Dananjay Srinivas, Maria Pacheco, Ashutosh Trivedi