Speaker: Grigory Yaroslavtsev

\nAffiliation: Indiana Un
iversity\, Bloomington

Title: Advances in Hierarchical Clustering
for Vector Data

\nAbstract:

\nCompared to the highly successful
flat clustering (e.g. k-means)\, despite its important role and applicatio
ns in data analysis\, hierarchical clustering has been lacking in rigorous
algorithmic studies until late due to absence of rigorous objectives. Sin
ce 2016\, a sequence of works has emerged and gave novel algorithms for th
is problem in the general metric setting. This was enabled by a breakthrou
gh by Dasgupta\, who introduced a formal objective into the study of hiera
rchical clustering.

In this talk I will give an overview of our re cent progress on models and scalable algorithms for hierarchical clusterin g applicable specifically to high-dimensional vector data. I will first di scuss various linkage-based algorithms (single-linkage\, average-linkage) and their formal properties with respect to various objectives. I will the n introduce a new projection-based approximation algorithm for vector data . The talk will be self-contained and doesn’t assume prior knowledge of cl ustering methods.

\nBased on joint works with Vadapalli (ICML’18) an d Charikar\, Chatziafratis and Niazadeh (AISTATS’19)

DTSTART;TZID=America/New_York:20190306T120000 DTEND;TZID=America/New_York:20190306T130000 SEQUENCE:0 SUMMARY:[Theory Seminar] Grigory Yaroslavtsev URL:https://www.cs.jhu.edu/~mdinitz/theory/event/theory-seminar-grigory-yar oslavtsev/ X-COST-TYPE:free END:VEVENT END:VCALENDAR