A neural network structure decomposition based on pruning and its visualization method

Atsushi Shibata; Jiajun Lu; Fangyan Dong; Kaoru Hirota

doi:10.20965/jaciii.2013.p0443

A neural network structure decomposition based on pruning and its visualization method

Atsushi Shibata, Jiajun Lu, Fangyan Dong, Kaoru Hirota

Tokyo Institute of Technology

Research output: Contribution to journal › Article › peer-review

Abstract

To decompose neural network structures for composite tasks, a pruning method and its visualization method are proposed. Visualization by placing the neurons on a 2D plane clarifies the structure related to each composited task. Experiments on a composite task using two tasks from a UCI dataset show that the neural network of the composite task contains more than 80% of neurons. The proposed methods target the transfer learning of robot motion, and results of an adaptation experiments are also referred.

Original language	English
Pages (from-to)	443-449
Number of pages	7
Journal	Journal of Advanced Computational Intelligence and Intelligent Informatics
Volume	17
Issue number	3
DOIs	https://doi.org/10.20965/jaciii.2013.p0443
Publication status	Published - May 2013
Externally published	Yes

Keywords

6 leg robot
Neural network
Pruning
Uci dataset
Visualization

Access to Document

10.20965/jaciii.2013.p0443

Cite this

@article{bafad2df4f7f4b0db88f9bfe5cc8c867,

title = "A neural network structure decomposition based on pruning and its visualization method",

abstract = "To decompose neural network structures for composite tasks, a pruning method and its visualization method are proposed. Visualization by placing the neurons on a 2D plane clarifies the structure related to each composited task. Experiments on a composite task using two tasks from a UCI dataset show that the neural network of the composite task contains more than 80% of neurons. The proposed methods target the transfer learning of robot motion, and results of an adaptation experiments are also referred.",

keywords = "6 leg robot, Neural network, Pruning, Uci dataset, Visualization",

author = "Atsushi Shibata and Jiajun Lu and Fangyan Dong and Kaoru Hirota",

year = "2013",

month = may,

doi = "10.20965/jaciii.2013.p0443",

language = "English",

volume = "17",

pages = "443--449",

journal = "Journal of Advanced Computational Intelligence and Intelligent Informatics",

issn = "1343-0130",

publisher = "Fuji Technology Press",

number = "3",

}

TY - JOUR

T1 - A neural network structure decomposition based on pruning and its visualization method

AU - Shibata, Atsushi

AU - Lu, Jiajun

AU - Dong, Fangyan

AU - Hirota, Kaoru

PY - 2013/5

Y1 - 2013/5

N2 - To decompose neural network structures for composite tasks, a pruning method and its visualization method are proposed. Visualization by placing the neurons on a 2D plane clarifies the structure related to each composited task. Experiments on a composite task using two tasks from a UCI dataset show that the neural network of the composite task contains more than 80% of neurons. The proposed methods target the transfer learning of robot motion, and results of an adaptation experiments are also referred.

AB - To decompose neural network structures for composite tasks, a pruning method and its visualization method are proposed. Visualization by placing the neurons on a 2D plane clarifies the structure related to each composited task. Experiments on a composite task using two tasks from a UCI dataset show that the neural network of the composite task contains more than 80% of neurons. The proposed methods target the transfer learning of robot motion, and results of an adaptation experiments are also referred.

KW - 6 leg robot

KW - Neural network

KW - Pruning

KW - Uci dataset

KW - Visualization

UR - http://www.scopus.com/inward/record.url?scp=84879396910&partnerID=8YFLogxK

U2 - 10.20965/jaciii.2013.p0443

DO - 10.20965/jaciii.2013.p0443

M3 - Article

AN - SCOPUS:84879396910

SN - 1343-0130

VL - 17

SP - 443

EP - 449

JO - Journal of Advanced Computational Intelligence and Intelligent Informatics

JF - Journal of Advanced Computational Intelligence and Intelligent Informatics

IS - 3

ER -

A neural network structure decomposition based on pruning and its visualization method

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this