We show that simultaneous perturbation can be used as an algorithm for on-line learning, and we report our theoretical investigation on generalization performance obtained with a statistical mechanical method. Asymptotic behavior of generalization error using this algorithm is on the order of t to the minus one-third power, where t is the learning time or the number of learning examples. This order is the same as that using well-known perceptron learning.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Seiji MIYOSHI, Hiroomi HIKAWA, Yutaka MAEDA, "Statistical Mechanical Analysis of Simultaneous Perturbation Learning" in IEICE TRANSACTIONS on Fundamentals,
vol. E92-A, no. 7, pp. 1743-1746, July 2009, doi: 10.1587/transfun.E92.A.1743.
Abstract: We show that simultaneous perturbation can be used as an algorithm for on-line learning, and we report our theoretical investigation on generalization performance obtained with a statistical mechanical method. Asymptotic behavior of generalization error using this algorithm is on the order of t to the minus one-third power, where t is the learning time or the number of learning examples. This order is the same as that using well-known perceptron learning.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E92.A.1743/_p
Copy
@ARTICLE{e92-a_7_1743,
author={Seiji MIYOSHI, Hiroomi HIKAWA, Yutaka MAEDA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Statistical Mechanical Analysis of Simultaneous Perturbation Learning},
year={2009},
volume={E92-A},
number={7},
pages={1743-1746},
abstract={We show that simultaneous perturbation can be used as an algorithm for on-line learning, and we report our theoretical investigation on generalization performance obtained with a statistical mechanical method. Asymptotic behavior of generalization error using this algorithm is on the order of t to the minus one-third power, where t is the learning time or the number of learning examples. This order is the same as that using well-known perceptron learning.},
keywords={},
doi={10.1587/transfun.E92.A.1743},
ISSN={1745-1337},
month={July},}
Copy
TY - JOUR
TI - Statistical Mechanical Analysis of Simultaneous Perturbation Learning
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1743
EP - 1746
AU - Seiji MIYOSHI
AU - Hiroomi HIKAWA
AU - Yutaka MAEDA
PY - 2009
DO - 10.1587/transfun.E92.A.1743
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E92-A
IS - 7
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - July 2009
AB - We show that simultaneous perturbation can be used as an algorithm for on-line learning, and we report our theoretical investigation on generalization performance obtained with a statistical mechanical method. Asymptotic behavior of generalization error using this algorithm is on the order of t to the minus one-third power, where t is the learning time or the number of learning examples. This order is the same as that using well-known perceptron learning.
ER -