Fixed-Size Pegasos for Large Scale Pinball Loss SVM thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Fixed-Size Pegasos for Large Scale Pinball Loss SVM

Published on Aug 26, 20133270 Views

Here we address the problem of learning pinball loss SVM model. We utilize modi fied Pegasos algorithm for large scale problems. The latter type of problems is leveraged for linearly non-separable cas

Related categories

Chapter list

Fixed-sized Pegasos for Large Scale Pinball Loss SVM00:00
Outline00:12
Stochastic programming01:09
Pegasos01:46
Pegasos cont'd02:02
Pinball loss03:03
Figures (1)04:28
Figures (2)05:03
Figures (3)05:28
Algorithm 105:54
Convergence bounds07:07
Fixed-size approach08:33
Algorithm 209:53
Toy datasets10:57
UCI datasets11:40
Convergence of pegasos algorithms12:24
Pegasos's "pros"13:04
References13:41
Thanks13:46