A VLSI-optimal constructive algorithm for classification problems

Sorin Draghici; Valeriu Beiu; Ishwar K. Sethi

A VLSI-optimal constructive algorithm for classification problems

Sorin Draghici, Valeriu Beiu, Ishwar K. Sethi

Research output: Contribution to conference › Paper › peer-review

Abstract

If neural networks are to be used on a large scale, they have to be implemented in hardware. However, the cost of the hardware implementation is critically sensitive to factors like the precision used for the weights, the total number of bits of information and the maximum fan-in used in the network. This paper presents a version of the Constraint Based Decomposition training algorithm which is able to produce networks using limited precision integer weights and units with limited fan-in. The algorithm is tested on the 2-spiral problem and the results are compared with other existing algorithms.

Original language	English
Pages	145-150
Number of pages	6
Publication status	Published - 1997
Externally published	Yes
Event	Proceedings of the 1997 Artificial Neural Networks in Engineering Conference, ANNIE'97 - St.Louis, MO, USA Duration: Nov 9 1997 → Nov 12 1997

Other

Other	Proceedings of the 1997 Artificial Neural Networks in Engineering Conference, ANNIE'97
City	St.Louis, MO, USA
Period	11/9/97 → 11/12/97

ASJC Scopus subject areas

Software

Cite this

@conference{76c28a9d8d6540189d9ef7412ba58a01,

title = "A VLSI-optimal constructive algorithm for classification problems",

abstract = "If neural networks are to be used on a large scale, they have to be implemented in hardware. However, the cost of the hardware implementation is critically sensitive to factors like the precision used for the weights, the total number of bits of information and the maximum fan-in used in the network. This paper presents a version of the Constraint Based Decomposition training algorithm which is able to produce networks using limited precision integer weights and units with limited fan-in. The algorithm is tested on the 2-spiral problem and the results are compared with other existing algorithms.",

author = "Sorin Draghici and Valeriu Beiu and Sethi, {Ishwar K.}",

year = "1997",

language = "English",

pages = "145--150",

note = "Proceedings of the 1997 Artificial Neural Networks in Engineering Conference, ANNIE'97 ; Conference date: 09-11-1997 Through 12-11-1997",

}

TY - CONF

T1 - A VLSI-optimal constructive algorithm for classification problems

AU - Draghici, Sorin

AU - Beiu, Valeriu

AU - Sethi, Ishwar K.

PY - 1997

Y1 - 1997

N2 - If neural networks are to be used on a large scale, they have to be implemented in hardware. However, the cost of the hardware implementation is critically sensitive to factors like the precision used for the weights, the total number of bits of information and the maximum fan-in used in the network. This paper presents a version of the Constraint Based Decomposition training algorithm which is able to produce networks using limited precision integer weights and units with limited fan-in. The algorithm is tested on the 2-spiral problem and the results are compared with other existing algorithms.

AB - If neural networks are to be used on a large scale, they have to be implemented in hardware. However, the cost of the hardware implementation is critically sensitive to factors like the precision used for the weights, the total number of bits of information and the maximum fan-in used in the network. This paper presents a version of the Constraint Based Decomposition training algorithm which is able to produce networks using limited precision integer weights and units with limited fan-in. The algorithm is tested on the 2-spiral problem and the results are compared with other existing algorithms.

UR - http://www.scopus.com/inward/record.url?scp=0031295420&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031295420&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0031295420

SP - 145

EP - 150

T2 - Proceedings of the 1997 Artificial Neural Networks in Engineering Conference, ANNIE'97

Y2 - 9 November 1997 through 12 November 1997

ER -

A VLSI-optimal constructive algorithm for classification problems

Abstract

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this