Machine Learning in Advertising at Yahoo!

27 Sep
Monday, 09/27/2010 1:00pm to 2:30pm

Abraham Bagherjeiran
Yahoo! Labs Silicon Valley

Computer Science Building, Room 150

Host: ACM Student Chapter

Finding people who are receptive to ads is one of the most challenging problems facing the internet age. When done badly, the world notices as people are inundated with poor-quality ads. When done well, a user is delighted as he or she finds a new product. Through their interactions with web pages, users give us very weak signals from which we want to gain insight into their purchasing interests. With machine learning, we transform these weak signals into a precise estimation of how valuable the user is to an advertiser. The advertiser can then show only the most appropriate ads to the users. In this talk we will discuss how machine learning is used in a new advertising system for display advertising at Yahoo!. During the talk, we will address several questions. How do we give value to the advertiser in mapping users to appropriate products? How can we scale cutting-edge research into a running system that evaluates billions of examples? How do we do all this without violating the user's privacy?

Bio: Abraham Bagherjeiran is a senior research scientist at Yahoo! in Santa Clara, California. He has been involved in advertising science research for over 3 years. He has published several papers in well-respected machine learning conferences some of which are related to advertising. Several products within Yahoo! have benefited from his work. His current areas of focus are machine learning, advertising, social network analysis, and programming in ruby.


Pizza will be provided.