Self-Supervised Learning for Object Recognition based on Kernel Discriminant-EM Algorithm

DMCA

Self-Supervised Learning for Object Recognition based on Kernel Discriminant-EM Algorithm (2001)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Ying Wu , Thomas S. Huang , Kentaro Toyama

Venue:	on Kernel Discriminant-EM Algorithm. ICCV 2001. 7.2.3 Related Vision Topics
Citations:	9 - 4 self

BibTeX

@INPROCEEDINGS{Wu01self-supervisedlearning,
    author = {Ying Wu and Thomas S. Huang and Kentaro Toyama},
    title = {Self-Supervised Learning for Object Recognition based on Kernel Discriminant-EM Algorithm},
    booktitle = {on Kernel Discriminant-EM Algorithm. ICCV 2001. 7.2.3 Related Vision Topics},
    year = {2001},
    pages = {275--280}
}

OpenURL

Abstract

In Proc. of IEEE Int'l Conf. on Computer Vision, Vancouver, Canada, 2001 It is often tedious and expensive to label large training data sets for learning-based object recognition systems. This problem could be alleviated by selfsupervised learning techniques, which take a hybrid of labeled and unlabeled training data to learn classifiers. Discriminant-EM (D-EM) proposed a framework for such tasks and current D-EM algorithm employed linear discriminant analysis. However, the algorithm is limited by its dependence on linear transformations. This paper extends the linear D-EM to nonlinear kernel algorithm, Kernel D-EM, based on kernel multiple discriminant analysis (KMDA). KMDA provides better ability to simplify the probabilistic structures of data distributions in a discrimination space. We propose two novel data-sampling schemes for e#cient training of kernel discriminants. Experimental results show that classifiers using KMDA learning compare with SVM performance on standard benchmark tests, and that Kernel D-EM outperforms a variety of supervised and semi-supervised learning algorithms for a hand-gesture recognition task and fingertip tracking task. 1

Keyphrases

self-supervised learning kernel discriminant-em algorithm object recognition kernel d-em unlabeled training data linear transformation standard benchmark test cient training kernel multiple discriminant analysis kernel discriminants ieee int linear discriminant analysis discrimination space computer vision hand-gesture recognition task learning-based object recognition system linear d-em novel data-sampling scheme svm performance data distribution semi-supervised learning algorithm data set probabilistic structure experimental result current d-em algorithm

DMCA