Computer Science team wins Best Paper Award at premier machine learning conference
An all-Yale team led by Computer Science faculty members Yang Cai and Manolis Zampetakis has been awarded the Best Paper Award at the 2025 Conference on Learning Theory (COLT), one of the most prestigious and selective conferences in machine learning.
The winning paper, titled "What Makes Treatment Effects Identifiable? Characterizations and Estimators Beyond Unconfoundedness," addresses a fundamental challenge in causal inference: determining when it’s possible to reliably estimate the effect of a treatment or intervention from observational data. The team developed a new unified condition that precisely characterizes when causal effects are identifiable. This condition extends beyond traditional assumptions, enabling the identification of treatment effects in complex scenarios that previous models could not handle, such as studies with deterministic treatment decisions.
Computer Science professors Yang Cai (left) and Manolis Zampetakis (right).
The research team includes Yang Cai (professor of computer science and economics), Manolis Zampetakis (assistant professor of computer science), postdoctoral associate Alkis Kalavasis, and graduate students Katerina Mamali and Anay Mehrotra. Kalavasis is a postdoc at the Yale Institute for Foundations of Data Science (FDS), while the rest of the team is from Yale's Computer Science Department. Their work bridges statistical learning theory and causal inference methodologies, opening new possibilities for analyzing observational studies with complex treatment mechanisms.
“This work helps researchers tackle numerous, previously unanswerable questions, like the impact of a scholarship program given only to students above a test score cutoff, that standard methods couldn’t reliably handle,” said Cai.
COLT is widely regarded as the flagship venue for theoretical machine learning research, known for its rigorous standards and selectivity. This recognition highlights Yale's leadership in foundational AI and machine learning research and demonstrates the impact of the university's theoretical computer science program in advancing the mathematical foundations of artificial intelligence.
More Details
Published Date
Jul 15, 2025
