09-recsys

I rating scale of user u values of other ratings user

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: r Q ,P ( u ,i )∈R ui − ( µ + bu + bi + q pu ) T i ) 2 goodness of fit ( + λ qi + pu + bu + bi 2 2 2 2 ) regularization Typically selected via gridsearch on a validation set Stochastic gradient decent to the rescue! 2/2/2011 Jure Leskovec, Stanford C246: Mining Massive Datasets 34 [Bellkor Team] Original model: rui = µ +bu + bi + qi pu Add time dependence to biases: rui = µ +bu(t)+ bi(t) +qi pu(t) Time-dependence parametrized by linear trends Add time dependence to user “factor weights” Models the fact that user’s interests over “genres” (the qs) may change over time Y. Koren, Collaborative filtering with temporal dynamics, KDD ’09 2/2/2011 Jure Leskovec, Stanford C246: Mining Massive Datasets 35 [Bellkor Team] Factor models: RMSE vs. #parameters 0.905 50 Basic SVD 100 … + What was Rated 200 0.900 … + Linear Time Factors … + Per-Day User Biases 50 0.895 RMSE 100 … + per-Day User Factors 200 50 0.890 100 0.885 100 200 200 500 500 50 100 0.880 200 500 1000 1500 0.875 10 100 1000 10000 100000 Millions of Parameters 2/2/2011 Jure Leskovec, Stanford C246: Mining Massive Datasets 36 2/2/2011 Jure Leskovec, Stanford C246: Mining Massive Datasets 37 BellKorPragmaticChaos BellKor: Yehuda Koren (now Yahoo!), Bob Bell, Chris Volinsky, AT&T BigChaos: Michael Jahrer, Andreas Toscher, 2 grad students from Austria Pragmatic Theory Martin Chabert, Martin Piotte, 2 engineers from Montreal June 26th submission triggers 30-day “last call” Submission timed purposely to coincide with vacation schedules 2/2/2011 Jure Leskovec, Stanford C246: Mining Massive Datasets 38 Ensemble team formed Group of other teams on leaderboard forms a new team Relies on combining their models Quickly also get a qualifying score over 10% BellKor Continue to eke out small improvements in their scores Realize that they are in direct competition with Ensemble Strategy Both teams carefully monitoring the leaderboard Only sure way to check for improvement is to submit a set of predictions This alerts the other team of your latest score 2/2/2011 Jure L...
View Full Document

This document was uploaded on 02/26/2014 for the course CS 246 at Stanford.

Ask a homework question - tutors are online