Algorithm for finding an interpretable simple neural network solution using PLS

Rasmus Bro

doi:10.1002/cem.1180090508

Algorithm for finding an interpretable simple neural network solution using PLS

Rasmus Bro^*

^*Corresponding author for this work

Food Analytics and Biotechnology

8 Citations (Scopus)

Abstract

This communication describes the combination of a feedforward neural network (NN) with one hidden neuron and partial least squares (PLS) regression. Through training of the neural network with an algorithm that is a combination of a modified simplex, PLS and certain numerical restrictions, one gains an NN solution that has several feasible properties: (i) as in PLS the solution is qualitatively interpretable; (ii) it works faster than or comparably with ordinary training algorithms for neural networks; (iii) it contains the linear solution as a limiting case. Another very important aspect of this training algorithm is the fact that outlier detection as in ordinary PLS is possible through loadings, scores and residuals. The algorithm is used on a simple non‐linear problem concerning fluorescence spectra of white sugar solutions.

Original language	English
Journal	Journal of Chemometrics
Volume	9
Issue number	5
Pages (from-to)	423-430
Number of pages	8
ISSN	0886-9383
DOIs	https://doi.org/10.1002/cem.1180090508
Publication status	Published - 1 Jan 1995

Keywords

interpretable
neural network
PLS
training

Access to Document

10.1002/cem.1180090508

Cite this

@article{b2e0e817b2694cf39b78fe1be8351ba9,

title = "Algorithm for finding an interpretable simple neural network solution using PLS",

abstract = "This communication describes the combination of a feedforward neural network (NN) with one hidden neuron and partial least squares (PLS) regression. Through training of the neural network with an algorithm that is a combination of a modified simplex, PLS and certain numerical restrictions, one gains an NN solution that has several feasible properties: (i) as in PLS the solution is qualitatively interpretable; (ii) it works faster than or comparably with ordinary training algorithms for neural networks; (iii) it contains the linear solution as a limiting case. Another very important aspect of this training algorithm is the fact that outlier detection as in ordinary PLS is possible through loadings, scores and residuals. The algorithm is used on a simple non‐linear problem concerning fluorescence spectra of white sugar solutions.",

keywords = "interpretable, neural network, PLS, training",

author = "Rasmus Bro",

year = "1995",

month = jan,

day = "1",

doi = "10.1002/cem.1180090508",

language = "English",

volume = "9",

pages = "423--430",

journal = "Journal of Chemometrics",

issn = "0886-9383",

publisher = "Wiley",

number = "5",

}

TY - JOUR

T1 - Algorithm for finding an interpretable simple neural network solution using PLS

AU - Bro, Rasmus

PY - 1995/1/1

Y1 - 1995/1/1

N2 - This communication describes the combination of a feedforward neural network (NN) with one hidden neuron and partial least squares (PLS) regression. Through training of the neural network with an algorithm that is a combination of a modified simplex, PLS and certain numerical restrictions, one gains an NN solution that has several feasible properties: (i) as in PLS the solution is qualitatively interpretable; (ii) it works faster than or comparably with ordinary training algorithms for neural networks; (iii) it contains the linear solution as a limiting case. Another very important aspect of this training algorithm is the fact that outlier detection as in ordinary PLS is possible through loadings, scores and residuals. The algorithm is used on a simple non‐linear problem concerning fluorescence spectra of white sugar solutions.

AB - This communication describes the combination of a feedforward neural network (NN) with one hidden neuron and partial least squares (PLS) regression. Through training of the neural network with an algorithm that is a combination of a modified simplex, PLS and certain numerical restrictions, one gains an NN solution that has several feasible properties: (i) as in PLS the solution is qualitatively interpretable; (ii) it works faster than or comparably with ordinary training algorithms for neural networks; (iii) it contains the linear solution as a limiting case. Another very important aspect of this training algorithm is the fact that outlier detection as in ordinary PLS is possible through loadings, scores and residuals. The algorithm is used on a simple non‐linear problem concerning fluorescence spectra of white sugar solutions.

KW - interpretable

KW - neural network

KW - PLS

KW - training

UR - http://www.scopus.com/inward/record.url?scp=84984373569&partnerID=8YFLogxK

U2 - 10.1002/cem.1180090508

DO - 10.1002/cem.1180090508

M3 - Journal article

AN - SCOPUS:84984373569

SN - 0886-9383

VL - 9

SP - 423

EP - 430

JO - Journal of Chemometrics

JF - Journal of Chemometrics

IS - 5

ER -

Algorithm for finding an interpretable simple neural network solution using PLS

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this