STA 290 Seminar: Cho-Jui Hsieh

STA 290 Seminar Series

DATE:             Thursday, June 1st, 2017, 4:10pm

LOCATION:    MSB 1147, Colloquium Room. Refreshments at 3:30pm in MSB 4110

SPEAKER:     Cho-Jui Hsieh, Statistics, UC Davis

TITLE:             “Learning from Comparisons”

ABSTRACT:   In this talk, we study several modern machine learning applications that involve learning from comparisons. In the first part of the talk, we discuss the problem of modeling higher order interactions in group comparisons, with applications in online MOBA games. In the second part of the talk, we develop a collaborative ranking algorithm for recommender systems, showing that by transforming rating data to pairwise comparisons, the proposed approach outperforms traditional matrix completion.

This is joint work with Yao Li, Liwei Wu, Minhao Cheng, Kevin Fuji, James Sharpnack and Fushing Hsieh.