George Montañez Wins Two IJCNN 2017 Awards -Machine Learning Department - Carnegie Mellon University

George Montañez Wins Two IJCNN 2017 Awards

George Montañez won Best Poster award and the INNS/Intel Best Student Paper award at the 2017 International Joint Conference on Neural Networks (IJCNN)

Both awards were for the same paper, "The LICORS Cabinet: Nonparametric Light Cone Methods for Spatio-Temporal Modeling," co-authored with Cosma Rohilla Shalizi.

The paper deals with the problem of modeling video and other high-dimensional spatio-temporal data. To make the problem tractable, local decompositions of structure (known as "light cones") are used, and from these probabilistic models are constructed. The models can be used for higher-level learning tasks, such as classification, regression, or even generating frames of video. The paper introduces three simple light-cone methods and applies the techniques to prediction tasks in semiconductor electron flow modeling and human video modeling. Possible future applications of light-cone methods include surveillance video anomaly detection and autonomous car video video-feed modeling.