SFASA Jan. 23
Wed. Seminar -- Netflix Recommendations: Data, Models, and Statistics
Title: |
Netflix Recommendations: Data, Models,
and Statistics |
Time: |
Wed. Jan. 23rd. 4:30pm -6:30pm (coffee/snack
from 4:30pm, presentation from 5:00pm) |
Speaker: |
Dr. Xavier Amatriain,
Director of Personalization Science and Engineering, Netfilx |
Location: |
San Jose State University,MacQuarrie Hall 520 (fifth floor) |
You may find a map of the campus at http://www.sjsu.edu/map/
MacQuarrie Hall is identified as MQH and is in
grid 2D on the map.
Parking: For on-campus
parking, one would probably park in the Òsouth parking garageÓ, also in grid 2D
on the map. Public parking would be on the second floor and higher.
Pay stations are located at each end of each floor, where one can purchase a
permit ($1/half hour, $8 max for the day).
Abstract:
Netflix is known for pushing the envelope of
recommendation technologies. In particular, the Netflix Prize put a focus on
using explicit user feedback to predict ratings using data mining and machine
learning techniques. Nowadays Netflix has moved to focusing on instant video
streaming over the internet and this has spurred
numerous changes in the way people use the service. In this talk I will give an
overview of the different techniques we currently use to personalize Netflix.
These methods include a great variety of approaches: From the Matrix
Factorization and Restricted Boltzmann Machines we learned during the Prize to
advanced Learning to Rank approaches. I will describe how we deal with the
different data sources and machine learning models and how we integrate them
into a flexible architecture that can deal with large offline batch jobs as
well as respond to real-time signals. Finally, I will talk about our
offline-online innovation cycle that connects our machine learning experiments
with the results of our AB tests, using data and statistics as our main
decision mechanism.
Speaker Bio:
Xavier Amatriain is Director
of Personalization Science and Engineering at Netfilx,
where he leads a team working on defining the next generation of personalized
experiences at Netflix. He is working on the cross-roads
of machine learning, software engineering, innovation, and agile methods.
Previous to this, he was a researcher focused on Recommender Systems and
neighboring areas such as Data Mining, Machine Learning, User Modeling, and
Social Networks. He has authored more than 50 papers in books, journals and
international conferences.
Return to www.sfasa.org