α-EM learning and its cookbook: From mixture-of-expert neural networks to movie random field

Yasuo Matsuyama; Takayuki Ikeda; Tomoaki Tanaka; Satoshi Furukawa; Naoki Takeda; Takeshi Niimoto

α-EM learning and its cookbook: From mixture-of-expert neural networks to movie random field

Yasuo Matsuyama^*, Takayuki Ikeda, Tomoaki Tanaka, Satoshi Furukawa, Naoki Takeda, Takeshi Niimoto

^*Corresponding author for this work

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

2 Citations (Scopus)

Abstract

The α-EM algorithm is a proper extension of the traditional log-EM algorithm. This new algorithm is based on the α-logarithm, while the traditional one uses the logarithm. The case of α = -1 corresponds to the log-EM algorithm. Since the speed of the α-EM algorithm was reported for learning problems, this paper shows that closed-form E-steps can be obtained for a wide class of problems. There is a set of common techniques. That is, a cookbook for the α-EM algorithm is presented. The recipes include unsupervised neural networks, supervised neural networks for various gating, hidden Markov models and Markov random fields for moving object segmentation. Reasoning for the speedup is also given.

Original language	English
Title of host publication	Proceedings of the International Joint Conference on Neural Networks
Place of Publication	United States
Publisher	IEEE
Pages	1368-1373
Number of pages	6
Volume	2
Publication status	Published - 1999
Event	International Joint Conference on Neural Networks (IJCNN'99) - Washington, DC, USA Duration: 1999 Jul 10 → 1999 Jul 16

Other

Other	International Joint Conference on Neural Networks (IJCNN'99)
City	Washington, DC, USA
Period	99/7/10 → 99/7/16

ASJC Scopus subject areas

Software

Cite this

Matsuyama, Y, Ikeda, T, Tanaka, T, Furukawa, S, Takeda, N & Niimoto, T 1999, α-EM learning and its cookbook: From mixture-of-expert neural networks to movie random field. in Proceedings of the International Joint Conference on Neural Networks. vol. 2, IEEE, United States, pp. 1368-1373, International Joint Conference on Neural Networks (IJCNN'99), Washington, DC, USA, 99/7/10.

@inproceedings{ad34347e92c14324a8903645c50a6e65,

title = "α-EM learning and its cookbook: From mixture-of-expert neural networks to movie random field",

abstract = "The α-EM algorithm is a proper extension of the traditional log-EM algorithm. This new algorithm is based on the α-logarithm, while the traditional one uses the logarithm. The case of α = -1 corresponds to the log-EM algorithm. Since the speed of the α-EM algorithm was reported for learning problems, this paper shows that closed-form E-steps can be obtained for a wide class of problems. There is a set of common techniques. That is, a cookbook for the α-EM algorithm is presented. The recipes include unsupervised neural networks, supervised neural networks for various gating, hidden Markov models and Markov random fields for moving object segmentation. Reasoning for the speedup is also given.",

author = "Yasuo Matsuyama and Takayuki Ikeda and Tomoaki Tanaka and Satoshi Furukawa and Naoki Takeda and Takeshi Niimoto",

year = "1999",

language = "English",

volume = "2",

pages = "1368--1373",

booktitle = "Proceedings of the International Joint Conference on Neural Networks",

publisher = "IEEE",

note = "International Joint Conference on Neural Networks (IJCNN'99) ; Conference date: 10-07-1999 Through 16-07-1999",

}

TY - GEN

T1 - α-EM learning and its cookbook

T2 - International Joint Conference on Neural Networks (IJCNN'99)

AU - Matsuyama, Yasuo

AU - Ikeda, Takayuki

AU - Tanaka, Tomoaki

AU - Furukawa, Satoshi

AU - Takeda, Naoki

AU - Niimoto, Takeshi

PY - 1999

Y1 - 1999

N2 - The α-EM algorithm is a proper extension of the traditional log-EM algorithm. This new algorithm is based on the α-logarithm, while the traditional one uses the logarithm. The case of α = -1 corresponds to the log-EM algorithm. Since the speed of the α-EM algorithm was reported for learning problems, this paper shows that closed-form E-steps can be obtained for a wide class of problems. There is a set of common techniques. That is, a cookbook for the α-EM algorithm is presented. The recipes include unsupervised neural networks, supervised neural networks for various gating, hidden Markov models and Markov random fields for moving object segmentation. Reasoning for the speedup is also given.

AB - The α-EM algorithm is a proper extension of the traditional log-EM algorithm. This new algorithm is based on the α-logarithm, while the traditional one uses the logarithm. The case of α = -1 corresponds to the log-EM algorithm. Since the speed of the α-EM algorithm was reported for learning problems, this paper shows that closed-form E-steps can be obtained for a wide class of problems. There is a set of common techniques. That is, a cookbook for the α-EM algorithm is presented. The recipes include unsupervised neural networks, supervised neural networks for various gating, hidden Markov models and Markov random fields for moving object segmentation. Reasoning for the speedup is also given.

UR - http://www.scopus.com/inward/record.url?scp=0033313112&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033313112&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0033313112

VL - 2

SP - 1368

EP - 1373

BT - Proceedings of the International Joint Conference on Neural Networks

PB - IEEE

CY - United States

Y2 - 10 July 1999 through 16 July 1999

ER -

α-EM learning and its cookbook: From mixture-of-expert neural networks to movie random field

Abstract

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this