Unification neural networks: unification by error-correction learning

Ekaterina Komendantskaya

doi:10.1093/jigpal/jzq012

Unification neural networks: unification by error-correction learning

Ekaterina Komendantskaya^*

^*Corresponding author for this work

School of Computer Science

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

Abstract

We show that the conventional first-order algorithm of unification can be simulated by finite artificial neural networks with one layer of neurons. In these unification neural networks, the unification algorithm is performed by error-correction learning. Each time-step of adaptation of the network corresponds to a single iteration of the unification algorithm. We present this result together with the library of learning functions and examples fully formalised in MATLAB Neural Network Toolbox.

Original language	English
Pages (from-to)	821-847
Number of pages	27
Journal	Logic Journal of the IGPL
Volume	19
Issue number	6
DOIs	https://doi.org/10.1093/jigpal/jzq012
Publication status	Published - Dec 2011

Keywords

Unification
Neuro-Symbolic Networks
Neural Network Learning
Error-correction Learning
Hybrid Networks
Connectionism
PROGRAMS

Access to Document

10.1093/jigpal/jzq012

Fellowship CLANN EP/F044046/1: Computational Logic in Artificial Neural Networks
Dyckhoff, R. (PI)
EPSRC
1/10/08 → 30/09/11
Project: Fellowship

Cite this

@article{020503c425604694949e0bbc78250970,

title = "Unification neural networks: unification by error-correction learning",

abstract = "We show that the conventional first-order algorithm of unification can be simulated by finite artificial neural networks with one layer of neurons. In these unification neural networks, the unification algorithm is performed by error-correction learning. Each time-step of adaptation of the network corresponds to a single iteration of the unification algorithm. We present this result together with the library of learning functions and examples fully formalised in MATLAB Neural Network Toolbox.",

keywords = "Unification, Neuro-Symbolic Networks, Neural Network Learning, Error-correction Learning, Hybrid Networks, Connectionism, PROGRAMS",

author = "Ekaterina Komendantskaya",

year = "2011",

month = dec,

doi = "10.1093/jigpal/jzq012",

language = "English",

volume = "19",

pages = "821--847",

journal = "Logic Journal of the IGPL",

issn = "1367-0751",

publisher = "Oxford University Press",

number = "6",

}

TY - JOUR

T1 - Unification neural networks

T2 - unification by error-correction learning

AU - Komendantskaya, Ekaterina

PY - 2011/12

Y1 - 2011/12

N2 - We show that the conventional first-order algorithm of unification can be simulated by finite artificial neural networks with one layer of neurons. In these unification neural networks, the unification algorithm is performed by error-correction learning. Each time-step of adaptation of the network corresponds to a single iteration of the unification algorithm. We present this result together with the library of learning functions and examples fully formalised in MATLAB Neural Network Toolbox.

AB - We show that the conventional first-order algorithm of unification can be simulated by finite artificial neural networks with one layer of neurons. In these unification neural networks, the unification algorithm is performed by error-correction learning. Each time-step of adaptation of the network corresponds to a single iteration of the unification algorithm. We present this result together with the library of learning functions and examples fully formalised in MATLAB Neural Network Toolbox.

KW - Unification

KW - Neuro-Symbolic Networks

KW - Neural Network Learning

KW - Error-correction Learning

KW - Hybrid Networks

KW - Connectionism

KW - PROGRAMS

U2 - 10.1093/jigpal/jzq012

DO - 10.1093/jigpal/jzq012

M3 - Article

SN - 1367-0751

VL - 19

SP - 821

EP - 847

JO - Logic Journal of the IGPL

JF - Logic Journal of the IGPL

IS - 6

ER -

Unification neural networks: unification by error-correction learning

Abstract

Keywords

Access to Document

Fingerprint

Projects

Fellowship CLANN EP/F044046/1: Computational Logic in Artificial Neural Networks

Cite this