One of the biggest challenges in building software is ensuring that what you build is correct. This year, at ICSE 2026, UMass LASER has two papers that bring us closer to that vision, using AI to automatically prove correctness.
If you're in Rio at ICSE this week, come see our talks. If not, read our papers: Cobblestone: https://t.co/NkwFp8x8Ue
ProofCoop: https://t.co/W71wb4anaw
The work identified what can go wrong when automated tools repair bugs and created an objective methodology for measuring patch quality used widely in modern repair research.
Original paper: https://t.co/9aBpwCz0HZ
https://t.co/rHjYfQgh7v
Incredibly honored to receive the ACM International Conference on the Foundations of Software Engineering 2025 Test of Time Award honorable mention for our work on overfitting in automated program repair. with Ted, @clegoues, & Earl.
LLMs offer an amazing opportunity to automate formal verification because the theorem prover is an oracle, identifying hallucinations and wrong proofs.
Hear about our Distinguished Paper Award work at @icseconf.bsky.social at 12:15 PM on Wed, room 212.
https://t.co/KQgjhyUMRU
Come learn how reinforcement learning can significantly improve automatic proof synthesis for formal verification!
Hear our talk on QEDCartographer at ICSE 2025 at 11:30 AM on Wednesday in room 212.
https://t.co/7lG17t9Z1F
Paper: https://t.co/69LGLEkLCQ
Contact Yuriy Brun <[email protected]> for questions and to discuss the position. See some of our latest work for examples of high-impact use of NLP for software correctness. https://t.co/b7WSmHb4j7 (ICSE'25) https://t.co/69LGLEljso (ICSE'25)
Are you graduating with a PhD? Do you work on program analysis, formal verification, software correctness, or AI? Apply to join the UMass LASER lab as a postdoctoral researcher, a vibrant team focused on using the latest NLP technology to ensure software correctness!
I am soliciting applications for a postdoctoral fellowship. If interested, please submit a CV, the names of 3 reference letter writers, and a short (~1 paragraph) statement of interest and a potential starting date by MARCH 15, 2025.
Do automatically generated patches help developers debug? Yes! Find out more in our ICSE'24 paper "Automated Program Repair, What Is It Good For? Not Absolutely Nothing!"
Video: https://t.co/5uNpXaLyj5
Paper: https://t.co/oZvZPuaKDK @clegoues@manningcics@ICSEconf
Thank you, IEEE Spectrum for the nice write up on our work. Research led by Emily First, with @MarkusNRabe and @TaliaRinger.
https://t.co/CHycxWkDlg
@manningcics
Our paper on automating formal verification (in Isabelle/HOL) won a Distinguished Paper Award at@FSEconf. Congrats to Emily First, @MarkusNRabe, @TaliaRinger! full paper: https://t.co/oK1gY4BYEP…@manningcics
@thecindyxiong@aimen_gaba@manningcics (2) describing system behavior using text rather than bar charts leads to people putting more weight on bias, and (3) explicitly labeling a system as biased has more effect than showing a history of biased behavior.
Zhanna Kaufman presenting our VIS'23 paper on how ML bias affects people's trust.
Paper: My Model is Unfair, Do People Even Care? Visual Design Affects Trust and Perceived Bias in Machine Learning
https://t.co/ytfVS2xZs6
@thecindyxiong , @aimen_gaba@manningcics
@thecindyxiong@aimen_gaba@manningcics Our paper aims to understand how bias and effectiveness of data-driven systems affect trust in systems. In a study with over 1,500 users, we find that (1) women weight bias more than men do when deciding whom to trust, regardless of whether the bias favors men or women,