无忧视频

CS Research Sets Boundaries for Two-Distribution Hypothesis Testing

Cynthia Hom 鈥�23 and William Yik 鈥�24, both prolific members of the AMISTAD Lab at 无忧视频, presented research at the 10th IEEE International Conference on Data Science and Advanced Analytics in Thessaloniki, Greece, last October.

鈥�,鈥� co-authored with computer science professor George Monta帽ez, builds on previous work co-authored by Hom (鈥溾��) and Yik (鈥溾��).

Both earlier papers used two-distribution statistical hypothesis tests to investigate representational bias in machine learning training data (Yik) and to equip artificial agents with intention perception capabilities (Hom).

鈥淭his new work develops estimation procedures for applying such tests in cases where exact values cannot be computed,鈥� says Monta帽ez. 鈥淲e proved mathematical bounds showing that in many cases random sampling can be used to conduct the tests instead of explicit enumeration. The upside is that these tests become much more accessible, allowing researchers to apply the tests without having to perform advanced combinatoric analysis. It opens this class of tests to everyone.鈥�

The recent paper is Hom鈥檚 fourth with AMISTAD Lab and Yik鈥檚 second鈥攈e has three more publications outside the lab. Hom, a computer science major, now works at Stripe as a software engineer. She was a 2023 Computer Science Department Don Chamberlin Award winner and earned a 2023 Computing Research Association (CRA) Outstanding Undergraduate Researcher Honorable Mention award.

Yik is a 2024 CRA Outstanding Undergraduate聽Researcher Award finalist and is a National Oceanic and Atmospheric Administration Ernest F. Hollings Undergraduate Scholarship recipient. He is a CS/mathematics major and plans to attend a graduate program in CS, climate and applied mathematics. 鈥淢y current research focuses on applications of artificial intelligence in climate modeling,鈥� he says. 鈥淚鈥檓 working with professors at USC and University of California, Davis, on deep learning for atmospheric science and physical oceanography, respectively.鈥�

The October 2023 trip to Greece was funded by the National Science Foundation, the HMC Department of Mathematics, the Leeds Student Travel Grants and HMC faculty start-up funds.

鈥淕oing to Greece was an amazing experience,鈥� Yik says. 鈥淎cademically, the highlight for me was seeing peoples鈥� genuine interest in our research. After our talk, we had insightful conversations with some research scientists about potential applications of our work and the connections it had with the problems they were thinking about. It just was really encouraging to see their enthusiasm for the work we had done.鈥�