Two Mini Talks on Clustering and Feature Selection
Cynthia Rudin, Duke, Computer Science and Electrical and Computer Engineering
I will present work in two mini-talks related to clustering in networks. In the first talk, there is a hidden network of crimes committed by the same individuals. In the second talk, the goal is to infer relationships between observations in machine learning. Both problems are (in a broad sense) subspace clustering problems. Mini Talk 1: Crime Series Detection via Subspace Clustering In crime series detection, the goal is to identify crimes that were committed by the same individuals. We cast this as a clustering problem with cluster-specific feature selection. It is joint work with my former student Tong Wang, and detectives Lt. Dan Wagner and Rich Sevieri of the Cambridge Police Department. Mini Talk 2: Bayesian Case Model The Bayesian Case Model (BCM) is a general framework for Bayesian case-based reasoning (CBR) and prototype classification and clustering. BCM brings the intuitive power of CBR to a Bayesian generative framework. The BCM learns prototypes, the “quintessential” observations that best represent clusters in a dataset, by performing joint inference on cluster labels, prototypes and important features. This is joint work with Been Kim and Julie Shah.
February, 6 2017 | 12:45 p.m. - 2:00 p.m. | Gross Hall 230E