Statistics Colloquium: STA 290
Note: this seminar is held jointly with Biostatistics
TUESDAY, December 3rd, 2013 at 4:10pm, MSB 1147 (Colloquium Room) Refreshments at 3:30pm in MSB 4110 (Statistics Lounge)
Speaker: DEEPAK AGARWAL, LinkedIn
Title: "Response Prediction for Computational Advertising Applications"
Abstract: In this talk I will provide an overview of various advertising models. Response prediction of a rare action like click is a fundamental input that is required to run these advertising systems in an automated fashion. This involves fitting large scale supervised learning algorithms to massive data consisting of hundreds of millions of observations and hundreds of thousands of predictors. I will describe ADMM algorithms and variations thereof that are easy to scale on readily available distributed computing infrastructures like Hadoop. Methods would be illustrated on LinkedIn Data.