Monday April 7, 12pm-1pm, 1116-E Klaus

My experiments with NMF applications

Ramakrishnan (Ramki) Kannan

Advisor: Prof. Haesun Park

ABSTRACT

The never ending debate - Does science drive applications or need is the mother of invention? My talk will support the first part of the debate about NMF driving new applications. In this talk, I will discuss my experiments with NMF applications in text domain, video and network data in the last 6 months. I will be sharing my experience and difficulties in using NMF in these applications such as scalability and interpretability. This will be chalk talk and will be using live examples. As an outcome of the talk, we can work together to find the support for "need driving the invention". :).

Most of the results in the talk are outcome of joint work with Prof. Park, Prof. Zha, Prof. Fujimoto, Dr. Hyenkyun woo, Dr. Charu Aggarwal, Yunlong He, Richard Boyd and Barry Drake.

BIO

Ramakrishnan (Ramki) Kannan is a Ph.D student advised by Prof. Park. He has 11 years of experience at IBM spanning across research, product development and services. He is an IBM Master Inventor. Currently, he is working on data mining and machine learning with specific focus on low rank approximation for special matrices. He completed his Master’s degree from Indian Institute of Science under the supervision of Prof. Narahari where he studied nash bargaining based bid optimization in sponsored search auctions.