full-length talk on the Netflix Prize competition

Jeff Howbert will be giving a talk on the Netflix Prize competition. It’s a full 60 minute talk with slides. He has presented this topic about 5-6 times locally, at Zillow and 3-4 other companies, and at the Puget Sound chapter of ASA.

The Netflix Prize: How a public contest and dedicated amateurs changed the science of recommender systems.

You must register at http://netflixprize.eventbrite.com .

Speaker:
Jeff Howbert, Research Scientist and Lecturer at University of Washington.

Abstract:
The Netflix Prize was the most widely followed machine learning competition in history. It offered a prize of $1,000,000 to anyone who could build a recommender system with an error rate at least 10% better than Netflix's internal system. Thousands of teams took part, and the contest lasted nearly three years. The speaker, a long-time active participant, will give an insider's perspective on two stories: 1) the machine learning approaches used during the competition, and how they substantially advanced the science of recommender systems, and 2) the unexpected emergence of two teams with winning scores, and their wild race to the finish. Part 2) is especially entertaining.

Agenda:
6:30 - 7:30 pm: 60-minute presentation
7:30 - 8:30 pm: Q&A and networking

Refreshments provided.