Please use this identifier to cite or link to this item: http://ktisis.cut.ac.cy/handle/10488/8209
Title: Infinite markov-switching maximum entropy discrimination machines
Authors: Chatzis, Sotirios P. 
Keywords: Bayesian nonparametrics
Dirichlet process models
Maximum entropy discrimination
Issue Date: 2013
Publisher: JMLR
Source: Journal of Machine Learning Research: Workshop and Conference Proceedings, 2013, Volume 28, Number 3, Pages 729-737
Abstract: In this paper, we present a method that combines the merits of Bayesian nonparametrics, specifically stick-breaking priors, and largemargin kernel machines in the context of sequential data classification. The proposed model employs a set of (theoretically) infinite interdependent large-margin classifiers as model components, that robustly capture local nonlinearity of complex data. The employed large-margin classifiers are connected in the context of a Markov-switching construction that allows for capturing complex temporal dynamics in the modeled datasets. Appropriate stick-breaking priors are imposed over the component switching mechanism of our model to allow for data-driven determination of the optimal number of component large-margin classifiers, under a standard nonparametric Bayesian inference scheme. Efficient model training is performed under the maximum entropy discrimination (MED) framework, which integrates the large-margin principle with Bayesian posterior inference. We evaluate our method using several real-world datasets, and compare it to state-of-the-art alternatives.
Description: Paper presented at 30th International Conference on Machine Learning, 2013, Atlanta, USA, 16 – 21 June.
URI: http://172.16.21.8/jspui/handle/10488/8209
ISSN: 1938-7228
Rights: © The author(s)
Appears in Collections:Άρθρα/Articles

Files in This Item:
File Description SizeFormat 
Chatzis.pdf2.88 MBAdobe PDFView/Open
Show full item record

Page view(s)

10
checked on Apr 25, 2017

Download(s) 20

3
checked on Apr 25, 2017

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.