Direct Synthesis of Neural Networks

Valeriu Beiu; John G. Taylor

doi:10.1109/MNNFS.1996.493800

Direct Synthesis of Neural Networks

Valeriu Beiu, John G. Taylor

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

Abstract

The paper overviews recent developments of a VLSI-friendly, constructive algorithm as well as detailing two extensions. The problem is to construct a neural network when m examples of n inputs are given (classification problem). The two extensions discussed are: (i) the use of analog comparators; and (ii) digital as well as analog solution to XOR-like problems. For a simple example (the two-spirals), we are able to show that the algorithm does a very “efficient” encoding of a given problem into the neural network it “builds”-when compared to the entropy of the given problem and to other learning algorithms. We are also able to estimate the number of bits needed to solve any classification problem for the general case. Being interested in the VLSI implementation of such networks, the optimum criteria are not only the classical size and depth, but also the connectivity and the number of bits for representing the weights-as such measures are closer estimates of the area and lead to better approximations of the AT^2.

Original language	English
Title of host publication	Direct Synthesis of Neural Networks
DOIs	https://doi.org/10.1109/MNNFS.1996.493800
Publication status	Published - Feb 12 1996
Event	Fifth International Conference on Microelectronics for Neural Networks - Lausanne, Switzerland Duration: Feb 12 1996 → …

Conference

Conference	Fifth International Conference on Microelectronics for Neural Networks
Period	2/12/96 → …

Access to Document

10.1109/MNNFS.1996.493800

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=493800&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D493800

Cite this

@inproceedings{abc68f06232a4302b669db8c6d031823,

title = "Direct Synthesis of Neural Networks",

abstract = "The paper overviews recent developments of a VLSI-friendly, constructive algorithm as well as detailing two extensions. The problem is to construct a neural network when m examples of n inputs are given (classification problem). The two extensions discussed are: (i) the use of analog comparators; and (ii) digital as well as analog solution to XOR-like problems. For a simple example (the two-spirals), we are able to show that the algorithm does a very “efficient” encoding of a given problem into the neural network it “builds”-when compared to the entropy of the given problem and to other learning algorithms. We are also able to estimate the number of bits needed to solve any classification problem for the general case. Being interested in the VLSI implementation of such networks, the optimum criteria are not only the classical size and depth, but also the connectivity and the number of bits for representing the weights-as such measures are closer estimates of the area and lead to better approximations of the AT^2.",

author = "Valeriu Beiu and Taylor, {John G.}",

year = "1996",

month = feb,

day = "12",

doi = "10.1109/MNNFS.1996.493800",

language = "English",

isbn = "0-8186-7373-7",

booktitle = "Direct Synthesis of Neural Networks",

note = "Fifth International Conference on Microelectronics for Neural Networks ; Conference date: 12-02-1996",

}

TY - GEN

T1 - Direct Synthesis of Neural Networks

AU - Beiu, Valeriu

AU - Taylor, John G.

PY - 1996/2/12

Y1 - 1996/2/12

N2 - The paper overviews recent developments of a VLSI-friendly, constructive algorithm as well as detailing two extensions. The problem is to construct a neural network when m examples of n inputs are given (classification problem). The two extensions discussed are: (i) the use of analog comparators; and (ii) digital as well as analog solution to XOR-like problems. For a simple example (the two-spirals), we are able to show that the algorithm does a very “efficient” encoding of a given problem into the neural network it “builds”-when compared to the entropy of the given problem and to other learning algorithms. We are also able to estimate the number of bits needed to solve any classification problem for the general case. Being interested in the VLSI implementation of such networks, the optimum criteria are not only the classical size and depth, but also the connectivity and the number of bits for representing the weights-as such measures are closer estimates of the area and lead to better approximations of the AT^2.

AB - The paper overviews recent developments of a VLSI-friendly, constructive algorithm as well as detailing two extensions. The problem is to construct a neural network when m examples of n inputs are given (classification problem). The two extensions discussed are: (i) the use of analog comparators; and (ii) digital as well as analog solution to XOR-like problems. For a simple example (the two-spirals), we are able to show that the algorithm does a very “efficient” encoding of a given problem into the neural network it “builds”-when compared to the entropy of the given problem and to other learning algorithms. We are also able to estimate the number of bits needed to solve any classification problem for the general case. Being interested in the VLSI implementation of such networks, the optimum criteria are not only the classical size and depth, but also the connectivity and the number of bits for representing the weights-as such measures are closer estimates of the area and lead to better approximations of the AT^2.

U2 - 10.1109/MNNFS.1996.493800

DO - 10.1109/MNNFS.1996.493800

M3 - Conference contribution

SN - 0-8186-7373-7

BT - Direct Synthesis of Neural Networks

T2 - Fifth International Conference on Microelectronics for Neural Networks

Y2 - 12 February 1996

ER -

Direct Synthesis of Neural Networks

Abstract

Conference

Access to Document

Fingerprint

Cite this