A Kernel Classifier for Distributions. Pozdnoukhov, A. & Bengio, S. Technical Report 05-32, IDIAP, 2005. Paper abstract bibtex This paper presents a new algorithm for classifying distributions. The algorithm combines the principle of margin maximization and a kernel trick, applied to distributions. Thus, it combines the discriminative power of support vector machines and the well-developed framework of generative models. It can be applied to a number of real-life tasks which include data represented as distributions. The algorithm can also be applied for introducing some prior knowledge on invariances into a discriminative model. We illustrate this approach in details for the case of Gaussian distributions, using a toy problem. We also present experiments devoted to the real-life problem of invariant image classification.
@techreport{pozdnoukhov:2005:05-32,
author = {A. Pozdnoukhov and S. Bengio},
title = {A Kernel Classifier for Distributions},
institution = {IDIAP},
year = 2005,
type = {Technical Report IDIAP-RR},
number = {05-32},
url = {publications/ps/rr05-32.ps.gz},
pdf = {publications/pdf/rr05-32.pdf},
djvu = {publications/djvu/rr05-32.djvu},
original = {2005/kernel_distribution_rr},
abstract = {This paper presents a new algorithm for classifying distributions. The algorithm combines the principle of margin maximization and a kernel trick, applied to distributions. Thus, it combines the discriminative power of support vector machines and the well-developed framework of generative models. It can be applied to a number of real-life tasks which include data represented as distributions. The algorithm can also be applied for introducing some prior knowledge on invariances into a discriminative model. We illustrate this approach in details for the case of Gaussian distributions, using a toy problem. We also present experiments devoted to the real-life problem of invariant image classification.},
categorie = {E},
}
Downloads: 0
{"_id":"3oWQB8a34go8g72oB","bibbaseid":"pozdnoukhov-bengio-akernelclassifierfordistributions-2005","authorIDs":[],"author_short":["Pozdnoukhov, A.","Bengio, S."],"bibdata":{"bibtype":"techreport","type":"Technical Report IDIAP-RR","author":[{"firstnames":["A."],"propositions":[],"lastnames":["Pozdnoukhov"],"suffixes":[]},{"firstnames":["S."],"propositions":[],"lastnames":["Bengio"],"suffixes":[]}],"title":"A Kernel Classifier for Distributions","institution":"IDIAP","year":"2005","number":"05-32","url":"publications/ps/rr05-32.ps.gz","pdf":"publications/pdf/rr05-32.pdf","djvu":"publications/djvu/rr05-32.djvu","original":"2005/kernel_distribution_rr","abstract":"This paper presents a new algorithm for classifying distributions. The algorithm combines the principle of margin maximization and a kernel trick, applied to distributions. Thus, it combines the discriminative power of support vector machines and the well-developed framework of generative models. It can be applied to a number of real-life tasks which include data represented as distributions. The algorithm can also be applied for introducing some prior knowledge on invariances into a discriminative model. We illustrate this approach in details for the case of Gaussian distributions, using a toy problem. We also present experiments devoted to the real-life problem of invariant image classification.","categorie":"E","bibtex":"@techreport{pozdnoukhov:2005:05-32,\n author = {A. Pozdnoukhov and S. Bengio},\n title = {A Kernel Classifier for Distributions},\n institution = {IDIAP},\n year = 2005,\n type = {Technical Report IDIAP-RR},\n number = {05-32},\n url = {publications/ps/rr05-32.ps.gz},\n pdf = {publications/pdf/rr05-32.pdf},\n djvu = {publications/djvu/rr05-32.djvu},\n original = {2005/kernel_distribution_rr},\n abstract = {This paper presents a new algorithm for classifying distributions. The algorithm combines the principle of margin maximization and a kernel trick, applied to distributions. Thus, it combines the discriminative power of support vector machines and the well-developed framework of generative models. It can be applied to a number of real-life tasks which include data represented as distributions. The algorithm can also be applied for introducing some prior knowledge on invariances into a discriminative model. We illustrate this approach in details for the case of Gaussian distributions, using a toy problem. We also present experiments devoted to the real-life problem of invariant image classification.},\n categorie = {E},\n}\n\n","author_short":["Pozdnoukhov, A.","Bengio, S."],"key":"pozdnoukhov:2005:05-32","id":"pozdnoukhov:2005:05-32","bibbaseid":"pozdnoukhov-bengio-akernelclassifierfordistributions-2005","role":"author","urls":{"Paper":"http://bengio.abracadoudou.com/publications/ps/rr05-32.ps.gz"},"downloads":0},"bibtype":"techreport","biburl":"http://bengio.abracadoudou.com/samy.bib","creationDate":"2020-03-18T03:43:27.509Z","downloads":0,"keywords":[],"search_terms":["kernel","classifier","distributions","pozdnoukhov","bengio"],"title":"A Kernel Classifier for Distributions","year":2005,"dataSources":["9NCW2CDr4M3s5DvNX"]}