Events Calendar

On Dimension Reduction and Clustering by Random Projections

Speaker: Teemu Roos, Helsinki Institute for Information Technology

Talk Title: On Dimension Reduction and Clustering by Random Projections

Abstract: I will discuss some interesting problems and promising solutions related to dimension reduction and clustering by random projections. It is well known that in high dimensional settings, random projections onto lower dimensional subspaces tends to retain pair-wise distances faithfully. Random projections are significantly faster that alternative techniques such as principal component analysis while providing comparable results. In the talk, I will present recent and ongoing work related to speeding up random projection-based computations by parallelization, fast approximate nearest neighbor searches using random projection trees, and applications in dimensional reduction and clustering.

Biography: Teemu Roos is an Assistant Professor at the Helsinki Institute for Information Technology HIIT and the Department of Computer Science, University of Helsinki, Finland. He received his MSc and PhD degrees in computer science from the same university in 2001 and 2007 respectively. Prof Roos's research interests include the theory and applications of probabilistic graphical models, information theory, and machine learning.

Host: Urbashi Mitra, ubli@usc.edu, EEB 536, x04667

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Gerrielyn Ramos