A Novel Approach for Efficient SVM Classification with Histogram Intersection Kernel thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

A Novel Approach for Efficient SVM Classification with Histogram Intersection Kernel

Published on Apr 03, 20144021 Views

The kernel trick – commonly used in machine learning and computer vision – enables learning of non-linear decision functions without having to explicitly map the original data to a high dimensional

Related categories

Chapter list

A Novel Approach for Efficient SVM Classification with Histogram Intersection Kernel00:00
Outline00:04
Histograms and SVM in Computer Vision - 100:20
Histograms and SVM in Computer Vision - 200:56
Context: SVM primal01:04
Context: SVM dual and non-linearity with kernel trick - 101:49
Context: SVM dual and non-linearity with kernel trick - 201:57
Context: SVM dual and non-linearity with kernel trick - 302:19
Context: SVM dual and non-linearity with kernel trick - 403:00
Context: SVM dual and non-linearity with kernel trick - 503:23
Problem addressed03:55
Fast Intersection Kernel SVM04:20
Explicit Feature Maps05:27
Proposed method: Recap06:15
Proposed method - 106:28
Proposed method: Pre-image of wΦ07:03
Proposed method - 207:34
Quasi convexity of objective08:09
Successive relaxation08:31
Stochastic sub-gradient based solver09:28
Experimental results09:51
Databases: Scene-1509:54
Databases: PASCAL VOC 200710:11
Implementation details10:31
Scene-15 database10:54
PASCAL VOC 2007 database11:16
Testing time11:27
Convergence for different parameter values12:09
Conclusion12:22
Thank you for your attention!13:00