Special Seminar: Shaofeng Jiang "New Coresets for Clustering: Beyond Euclidean Geometry"

New Coresets for Clustering: Beyond Euclidean Geometry

Shaofeng Jiang

Abstract: k-Clustering (e.g. k-median/k-means) is a basic task for data analysis and machine learning. However, classic clustering algorithms do not scale well on huge data sets. To this end, coresets were introduced as a powerful data reduction technique that turns a huge dataset into a tiny proxy. Moreover, coresets have been successfully applied to clustering in various settings including streaming and distributed computing.

Coresets for k-clustering in Euclidean spaces have been very well studied. However, very few results are known when the space is beyond Euclidean or the objective is more general than k-clustering. In this talk, I will introduce a series of my recent works on coresets, including coresets for k-clustering in doubling spaces, in planar graphs, and generalized coresets for flexible and fair clustering.

Bio: Shaofeng Jiang is currently a postdoctoral researcher in the Weizmann Institute of Science hosted by Robert Krauthgamer. Before joining Weizmann, he obtained his Ph.D. degree from the University of Hong Kong at 2017. His research interest is generally theoretical computer science, and especially algorithms for massive data sets, approximation algorithms and online algorithms. He is a recipient of an MSRA Fellowship Nomination Award, and an Outstanding Achievements in Postdoctoral Research Prize at the Weizmann Institute of Science.