Rare Category Detection Using Semi-Supervised Active Learning

Dr. Rohan Loveland will present the research talk, "Finding Needles in Haystacks: Rare Category Detection Using Semi-Supervised Active Learning."

Description: Rare category detection addresses the problem of exploring data sets that are too large for unaided analysis. The Farpoint algorithm utilizes machine learning techniques to discover sparsely represented classes, through interactive queries and semi-supervised clustering.

Application of the algorithm to skewed MNIST datasets is used to empirically characterize performance.