Learning to Compare using Operator-Valued Large-Margin Classifiers thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Learning to Compare using Operator-Valued Large-Margin Classifiers

Published on Feb 25, 20073486 Views

The proposed method uses homonymous and heteronymous examplepairs to train a linear preprocessor on a kernel-induced Hilbert space. The algorithm seeks to optimize the expected performance of elementa

Related categories

Chapter list

a binary classification task for pairs00:00
a binary classification task for pairs0101:18
a binary classification task for pairs0202:28
a binary classification task for pairs0302:45
pair classifiers induced by linear transformations03:15
pair classifiers induced by linear transformations0104:49
pair classifiers induced by linear transformations0205:42
estimation and generalization07:45
estimation and generalization0109:27
regularized objectives11:45
regularized objectives0112:53
optimization problem14:25
optimization problem0114:41
optimization problem0215:05
algorithm15:30
experiments15:46
rotation- and scale-invariant character recognition16:33
rotation- and scale-invariant character recognition0116:56
rotation- and scale-invariant character recognition0217:05
results for rotation/scale-invariant OCR17:13