INFO 6150 - Advanced Topic Modeling

(crosslisted) CS 6788

Fall. 3 credits. Letter grades only.

Prerequisite: familiarity with Bayesian statistics and probabilistic modeling. Permission of instructor required. Enrollment limited to: graduate students or seniors.

D. Mimno.

Statistical topic models such as LDA provide a powerful tool for discovering themes in large unlabeled text corpora. They are increasingly popular in a wide range of fields, both as a data-driven alternative to manual document coding methods, and also as an example of a difficult but tractable problem in statistical inference. This course will cover Bayesian model construction, inference techniques, evaluation, and applications beyond text such as community detection in networks and population admixture in genetics.