Invited Talk by Chetan Verma on Automated Web Video Classification
Title of the talk: Automated Web Video Classification
Date &Time:Monday,18 Nov 2013 10:30-11:30
Venue: Room No-123
Personalization applications such as content recommendation, product recommendation and advertisements, and social network related recommendations, can be quite beneficial for both, service providers and the end users. Such applications need to understand user preferences in order to provide customized services. A commonly used approach for understanding user preferences is through classification of the viewed content - such as of images, videos, webpages. Since the categories of interest for various applications are different, labeled training data to train classification models for appropriate set of categories is usually unavailable. In this talk, we will focus on the problem of text-based multi-class single-label classification of web video webpages (such as YouTube). We study the feasibility and effectiveness of a fully automated framework to obtain training videos to enable classification of web videos to any arbitrary set of categories, as desired by the personalization application. We investigate the desired properties in training data that leads to high performance of the trained classification models. We then develop an approach to identify and score keywords based on their suitability to retrieve training videos with the desired properties, for the specified set of categories. YouTube based experimental results are discussed.
Chetan is a PhD student at University of California San Diego (UCSD). His broad areas of focus are data mining, machine learning, text mining. Prior to joining UCSD, he worked with Qualcomm, before which he did his bachelors from IIT Madras. He has worked with Samsung Labs, and is currently working in collaboration with Yahoo Labs.