Transfer Learning in Collaborative Filtering

PhD Thesis Proposal Defence

Title: "Transfer Learning in Collaborative Filtering"


Mr. Weike Pan


Transfer learning and collaborative filtering have been studied in
each community separately since early 1990s and were married in late
2000s. Collaborative filtering is a major intelligent component in
various recommender systems, like movie recommendation in Netflix,
people recommendation in Tencent Weibo (microblog), and advertisement
recommendation in Facebook. Transfer learning in collaborative
filtering (TLCF) is studied to address the data sparsity problem in
the user-item preference data in recommender systems. In this
proposal, we aim to develop this new multidisciplinary area in several
aspects. First, we survey transfer learning works w.r.t model-based
transfer, instance-based transfer and feature-based transfer, and
collaborative filtering works w.r.t. memory-based methods and
model-based methods. Second, we summarize related TLCF works proposed
in the background of both transfer learning and non transfer learning
from four dimensions of auxiliary data, content, context, network and
feedback. Third, we study four new TLCF problem settings for movie
recommendation and people recommendation, and propose a general
learning framework and four novel solutions correspondingly. Finally,
we make a link between traditional transfer learning and TLCF from a
unified view of model-based transfer, instance-based transfer, and
feature-based transfer. Some preliminary results and thesis progress
are also included.

Date:                   Tuesday, 17 January 2012

Time:                   1:00pm - 3:00pm

Venue:                  Room 3501
                         lifts 25/26

Committee Members:      Prof. Qiang Yang (Supervisor)
                         Dr. Sunghun Kim (Chairperson)
 			Prof. Dit-Yan Yeung
 			Prof. Nevin Zhang

**** ALL are Welcome ****