Analysis for characteristics of GA-based learning method of binary neural networks

T Hirane; T Toryu; H Nakano; A Miyauchi; Hidehiro NAKANO

doi:10.1007/11550822_51

Analysis for characteristics of GA-based learning method of binary neural networks

T Hirane, T Toryu, H Nakano, A Miyauchi, Hidehiro NAKANO

Department of Computer Science

Research output: Contribution to journal › Misc

Abstract

In this paper, we analyze characteristics of GA-based learning method of Binary Neural Networks (BNN). First, we consider coding methods to a chromosome in a GA and discuss the necessary chromosome length for a learning of BNN. Then, we compare some selection methods in a GA. We show that the learning results can be obtained in the less number of generations by properly setting selection methods and parameters in a GA. We also show that the quality of the learning results can be almost the same as that of the conventional method. These results can be verified by numerical experiments.

Translated title of the contribution	Analysis for characteristics of GA-based learning method of binary neural networks
Original language	American English
Pages (from-to)	323 - 329
Journal	ARTIFICIAL NEURAL NETWORKS: BIOLOGICAL INSPIRATIONS - ICANN 2005, PT 1, PROCEEDINGS
Volume	3696
DOIs	https://doi.org/10.1007/11550822_51
State	Published - 2005

Access to Document

10.1007/11550822_51

Cite this

@article{443b1015323643c1880586bae068f721,

title = "Analysis for characteristics of GA-based learning method of binary neural networks",

abstract = "In this paper, we analyze characteristics of GA-based learning method of Binary Neural Networks (BNN). First, we consider coding methods to a chromosome in a GA and discuss the necessary chromosome length for a learning of BNN. Then, we compare some selection methods in a GA. We show that the learning results can be obtained in the less number of generations by properly setting selection methods and parameters in a GA. We also show that the quality of the learning results can be almost the same as that of the conventional method. These results can be verified by numerical experiments.",

author = "\{T Hirane\} and \{T Toryu\} and \{H Nakano\} and \{A Miyauchi\} and Hidehiro NAKANO",

year = "2005",

doi = "10.1007/11550822\_51",

language = "American English",

volume = "3696",

pages = "323 -- 329",

journal = "ARTIFICIAL NEURAL NETWORKS: BIOLOGICAL INSPIRATIONS - ICANN 2005, PT 1, PROCEEDINGS",

issn = "0302-9743",

publisher = "SPRINGER-VERLAG BERLIN",

}

TY - JOUR

T1 - Analysis for characteristics of GA-based learning method of binary neural networks

AU - T Hirane, null

AU - T Toryu, null

AU - H Nakano, null

AU - A Miyauchi, null

AU - NAKANO, Hidehiro

PY - 2005

Y1 - 2005

N2 - In this paper, we analyze characteristics of GA-based learning method of Binary Neural Networks (BNN). First, we consider coding methods to a chromosome in a GA and discuss the necessary chromosome length for a learning of BNN. Then, we compare some selection methods in a GA. We show that the learning results can be obtained in the less number of generations by properly setting selection methods and parameters in a GA. We also show that the quality of the learning results can be almost the same as that of the conventional method. These results can be verified by numerical experiments.

AB - In this paper, we analyze characteristics of GA-based learning method of Binary Neural Networks (BNN). First, we consider coding methods to a chromosome in a GA and discuss the necessary chromosome length for a learning of BNN. Then, we compare some selection methods in a GA. We show that the learning results can be obtained in the less number of generations by properly setting selection methods and parameters in a GA. We also show that the quality of the learning results can be almost the same as that of the conventional method. These results can be verified by numerical experiments.

U2 - 10.1007/11550822_51

DO - 10.1007/11550822_51

M3 - Misc

SN - 0302-9743

VL - 3696

SP - 323

EP - 329

JO - ARTIFICIAL NEURAL NETWORKS: BIOLOGICAL INSPIRATIONS - ICANN 2005, PT 1, PROCEEDINGS

JF - ARTIFICIAL NEURAL NETWORKS: BIOLOGICAL INSPIRATIONS - ICANN 2005, PT 1, PROCEEDINGS

ER -

Analysis for characteristics of GA-based learning method of binary neural networks

Abstract

Access to Document

Fingerprint

Cite this