Training neural networks using taguchi methods: Overcoming interaction problems

Alagappan Viswanathan; Christopher MacLeod; Grant Maxwell; Sashank Kalidindi

Training neural networks using taguchi methods: Overcoming interaction problems

Alagappan Viswanathan, Christopher MacLeod, Grant Maxwell, Sashank Kalidindi

UHI North West and Hebrides

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

3 Citations (Scopus)

Abstract

Taguchi Methods (and other orthogonal arrays) may be used to train small Artificial Neural Networks very quickly in a variety of tasks. These include, importantly, Control Systems. Previous experimental work has shown that they could be successfully used to train single layer networks with no difficulty. However, interaction between layers precluded the successful reliable training of multi-layered networks. This paper describes a number of successful strategies which may be used to overcome this problem and demonstrates the ability of such networks to learn non-linear mappings.

Original language	English
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages	103-108
Number of pages	6
Publication status	Published - 2005
Event	15th International Conference on Artificial Neural Networks: Biological Inspirations - ICANN 2005 - Warsaw, Poland Duration: 11 Sept 2005 → 15 Sept 2005

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	3697 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	15th International Conference on Artificial Neural Networks: Biological Inspirations - ICANN 2005
Country/Territory	Poland
City	Warsaw
Period	11/09/05 → 15/09/05

Handle.net

Persistent link

Cite this

Viswanathan, A., MacLeod, C., Maxwell, G., & Kalidindi, S. (2005). Training neural networks using taguchi methods: Overcoming interaction problems. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 103-108). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3697 LNCS).

Viswanathan, Alagappan ; MacLeod, Christopher ; Maxwell, Grant et al. / Training neural networks using taguchi methods : Overcoming interaction problems. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2005. pp. 103-108 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a5a6994fa8ff41feb7b0af9f794c1fd1,

title = "Training neural networks using taguchi methods: Overcoming interaction problems",

abstract = "Taguchi Methods (and other orthogonal arrays) may be used to train small Artificial Neural Networks very quickly in a variety of tasks. These include, importantly, Control Systems. Previous experimental work has shown that they could be successfully used to train single layer networks with no difficulty. However, interaction between layers precluded the successful reliable training of multi-layered networks. This paper describes a number of successful strategies which may be used to overcome this problem and demonstrates the ability of such networks to learn non-linear mappings.",

author = "Alagappan Viswanathan and Christopher MacLeod and Grant Maxwell and Sashank Kalidindi",

year = "2005",

language = "English",

isbn = "3540287558",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "103--108",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

note = "15th International Conference on Artificial Neural Networks: Biological Inspirations - ICANN 2005 ; Conference date: 11-09-2005 Through 15-09-2005",

}

Viswanathan, A, MacLeod, C, Maxwell, G & Kalidindi, S 2005, Training neural networks using taguchi methods: Overcoming interaction problems. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3697 LNCS, pp. 103-108, 15th International Conference on Artificial Neural Networks: Biological Inspirations - ICANN 2005, Warsaw, Poland, 11/09/05.

Training neural networks using taguchi methods: Overcoming interaction problems. / Viswanathan, Alagappan; MacLeod, Christopher; Maxwell, Grant et al.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2005. p. 103-108 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3697 LNCS).

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

TY - GEN

T1 - Training neural networks using taguchi methods

T2 - 15th International Conference on Artificial Neural Networks: Biological Inspirations - ICANN 2005

AU - Viswanathan, Alagappan

AU - MacLeod, Christopher

AU - Maxwell, Grant

AU - Kalidindi, Sashank

PY - 2005

Y1 - 2005

N2 - Taguchi Methods (and other orthogonal arrays) may be used to train small Artificial Neural Networks very quickly in a variety of tasks. These include, importantly, Control Systems. Previous experimental work has shown that they could be successfully used to train single layer networks with no difficulty. However, interaction between layers precluded the successful reliable training of multi-layered networks. This paper describes a number of successful strategies which may be used to overcome this problem and demonstrates the ability of such networks to learn non-linear mappings.

AB - Taguchi Methods (and other orthogonal arrays) may be used to train small Artificial Neural Networks very quickly in a variety of tasks. These include, importantly, Control Systems. Previous experimental work has shown that they could be successfully used to train single layer networks with no difficulty. However, interaction between layers precluded the successful reliable training of multi-layered networks. This paper describes a number of successful strategies which may be used to overcome this problem and demonstrates the ability of such networks to learn non-linear mappings.

UR - http://www.scopus.com/inward/record.url?scp=33646242029&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33646242029&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:33646242029

SN - 3540287558

SN - 9783540287551

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 103

EP - 108

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Y2 - 11 September 2005 through 15 September 2005

ER -

Viswanathan A, MacLeod C, Maxwell G, Kalidindi S. Training neural networks using taguchi methods: Overcoming interaction problems. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2005. p. 103-108. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

Training neural networks using taguchi methods: Overcoming interaction problems

Abstract

Publication series

Conference

Handle.net

Other files and links

Fingerprint

Cite this