Non-logarithmic information measures, α-weighted em algorithms and speedup of learning

Y. Matsuyama

doi:10.1109/ISIT.1998.708990

Non-logarithmic information measures, α-weighted em algorithms and speedup of learning

Y. Matsuyama^*

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

3 Citations (Scopus)

Abstract

Starting from Renyi's α-divergence, a class of generalized EM algorithms called the α-EM algorithms of the WEM algorithms are derived. Merits of this generalization are found on speedup of learning, i.e., acceleration of convergence. Discussions include novel α-versions of logarithm, efficient scores, information matrices and the Cramer-Rao bound. The speedup is examined on Gaussian mixture learning systems.

Original language	English
Title of host publication	IEEE International Symposium on Information Theory - Proceedings
Pages	385
Number of pages	1
DOIs	https://doi.org/10.1109/ISIT.1998.708990
Publication status	Published - 1998
Event	1998 IEEE International Symposium on Information Theory, ISIT 1998 - Cambridge, MA Duration: 1998 Aug 16 → 1998 Aug 21

Other

Other	1998 IEEE International Symposium on Information Theory, ISIT 1998
City	Cambridge, MA
Period	98/8/16 → 98/8/21

ASJC Scopus subject areas

Applied Mathematics
Modelling and Simulation
Theoretical Computer Science
Information Systems

Access to Document

10.1109/ISIT.1998.708990

Cite this

@inproceedings{2d4764f51f0b4615a41da6dc0577b623,

title = "Non-logarithmic information measures, α-weighted em algorithms and speedup of learning",

abstract = "Starting from Renyi's α-divergence, a class of generalized EM algorithms called the α-EM algorithms of the WEM algorithms are derived. Merits of this generalization are found on speedup of learning, i.e., acceleration of convergence. Discussions include novel α-versions of logarithm, efficient scores, information matrices and the Cramer-Rao bound. The speedup is examined on Gaussian mixture learning systems.",

author = "Y. Matsuyama",

year = "1998",

doi = "10.1109/ISIT.1998.708990",

language = "English",

isbn = "0780350006",

pages = "385",

booktitle = "IEEE International Symposium on Information Theory - Proceedings",

note = "1998 IEEE International Symposium on Information Theory, ISIT 1998 ; Conference date: 16-08-1998 Through 21-08-1998",

}

TY - GEN

T1 - Non-logarithmic information measures, α-weighted em algorithms and speedup of learning

AU - Matsuyama, Y.

PY - 1998

Y1 - 1998

N2 - Starting from Renyi's α-divergence, a class of generalized EM algorithms called the α-EM algorithms of the WEM algorithms are derived. Merits of this generalization are found on speedup of learning, i.e., acceleration of convergence. Discussions include novel α-versions of logarithm, efficient scores, information matrices and the Cramer-Rao bound. The speedup is examined on Gaussian mixture learning systems.

AB - Starting from Renyi's α-divergence, a class of generalized EM algorithms called the α-EM algorithms of the WEM algorithms are derived. Merits of this generalization are found on speedup of learning, i.e., acceleration of convergence. Discussions include novel α-versions of logarithm, efficient scores, information matrices and the Cramer-Rao bound. The speedup is examined on Gaussian mixture learning systems.

UR - http://www.scopus.com/inward/record.url?scp=84871368049&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84871368049&partnerID=8YFLogxK

U2 - 10.1109/ISIT.1998.708990

DO - 10.1109/ISIT.1998.708990

M3 - Conference contribution

AN - SCOPUS:84871368049

SN - 0780350006

SN - 9780780350007

SP - 385

BT - IEEE International Symposium on Information Theory - Proceedings

T2 - 1998 IEEE International Symposium on Information Theory, ISIT 1998

Y2 - 16 August 1998 through 21 August 1998

ER -

Non-logarithmic information measures, α-weighted em algorithms and speedup of learning

Abstract

Other

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this