Collaborative Filtering with the Trace Norm: Learning, Bounding, and Transducing  thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Collaborative Filtering with the Trace Norm: Learning, Bounding, and Transducing

Published on Aug 02, 20113125 Views

Trace-norm regularization is a widely-used and successful approach for collaborative filtering and matrix completion. However, its theoretical understanding is surprisingly weak, and despite previo

Related categories

Chapter list

Collaborative Filtering with the Trace Norm: Learning, Bounding and Transducing00:00
Collaborative Filtering and Matrix Completion00:03
Trace-Norm Regularization00:44
Bottom Lines02:03
Sample Complexity03:21
Sample Complexity –Uniform Distribution04:20
Sample Complexity –General Distributions05:03
Tightness05:48
But This is Fishy ...06:38
Analysis07:24
Boundedness - 108:01
Boundedness - 208:56
Proof Idea09:23
Transductive Learning - 110:24
Transductive Learning - 211:10
Transductive Learning - 311:19
Transductive Learning - 411:22
Transductive Learning - 511:52
Transductive Learning - 611:53
Transductive Learning - 712:30
Does Enforcing Boundedness Help?14:15
Conclusions15:53
Sample Complexity of Trace-norm? - 116:02
Sample Complexity of Trace-norm? - 218:31
Sample Complexity of Trace-norm? - 318:55
Sample Complexity of Trace-norm? - 419:05
Sample Complexity of Trace-norm? - 519:31