Comparison of Fisher's linear discriminant to multilayer perceptron networks in the classification of vapors using sensor array data

Matteo Pardo, Brian C. Sisk, Giorgio Sberveglieri, Nathan S Lewis

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Two different classification methods, Fisher's linear discriminant (FLD) and a multilayer perceptron neural network (MLP), were directly compared with respect to their abilities to differentiate response patterns arising from arrays of chemical vapor detectors. The algorithms were compared in five different types of tasks that had been selected because they produced classification problems of varying character and difficulty. In one task, an array of 20 compositionally distinct carbon black-polymer composite vapor detectors was exposed to P/P0 = 0.0075 1-propanol and P/P0 = 0.0083 2-propanol, where P and P0 are the partial pressure and standard vapor pressure, respectively, of a given analyte. The second task consisted of classification of a mixture of P/P0 = 0.011 1-propanol and P/P0 = 0.0090 2-propanol versus a mixture of P/P0 = 0.0090 1-propanol and P/P0 = 0.011 2-propanol. A third task consisted of multiple concentrations of three hydrocarbons, and a fourth task involved clustering two hydrocarbons in the presence of a variable background composition. An additional dataset was generated by exposing an array of five thin-film metal-oxide sensors to the headspace of seven different coffee blends. In each case, the MLP and FLD techniques were compared using the 5-sensor subset of the 20 available sensors that proved optimal for that dataset. The FLD and MLP algorithms yielded comparable performance on straightforward classification tasks, whereas the MLP technique yielded better performance on tasks that involved non-linear classification boundaries. In addition, for the four datasets produced by the carbon black-polymer composite detector array, the performance of each possible 5-sensor subset was evaluated using both signal processing approaches. The performance of the best 5-sensor subset selected with MLP was found to be slightly better than the performance of the FLD-selected subsets, and the performance of the median 5-sensor subset using MLP was nearer to that of the optimal subset than the median sensor array selected by FLD. In one case, the optimal test set performance distribution was found to be significantly better with MLP than with FLD: MLP had a clear advantage (86% versus 57% correct classification rate) when applied to the "coffees" dataset, and this trend is likely applicable to other multi-cluster classification tasks that consisted of non-Gaussian shaped data in lower-dimensional spaces.

Original languageEnglish
Pages (from-to)647-655
Number of pages9
JournalSensors and Actuators, B: Chemical
Volume115
Issue number2
DOIs
Publication statusPublished - Jun 26 2006

Fingerprint

self organizing systems
Sensor arrays
Multilayer neural networks
Propanol
Vapors
vapors
set theory
sensors
1-Propanol
2-Propanol
Sensors
Soot
Coffee
coffee
Hydrocarbons
Carbon black
Detectors
Polymers
detectors
hydrocarbons

Keywords

  • Artificial neural networks
  • Fisher's linear discriminant
  • Multilayer perceptrons
  • Vapor sensors

ASJC Scopus subject areas

  • Analytical Chemistry
  • Electrochemistry
  • Electrical and Electronic Engineering

Cite this

Comparison of Fisher's linear discriminant to multilayer perceptron networks in the classification of vapors using sensor array data. / Pardo, Matteo; Sisk, Brian C.; Sberveglieri, Giorgio; Lewis, Nathan S.

In: Sensors and Actuators, B: Chemical, Vol. 115, No. 2, 26.06.2006, p. 647-655.

Research output: Contribution to journalArticle

@article{6ed8affa53c04fe7ba9137c93ae2802f,
title = "Comparison of Fisher's linear discriminant to multilayer perceptron networks in the classification of vapors using sensor array data",
abstract = "Two different classification methods, Fisher's linear discriminant (FLD) and a multilayer perceptron neural network (MLP), were directly compared with respect to their abilities to differentiate response patterns arising from arrays of chemical vapor detectors. The algorithms were compared in five different types of tasks that had been selected because they produced classification problems of varying character and difficulty. In one task, an array of 20 compositionally distinct carbon black-polymer composite vapor detectors was exposed to P/P0 = 0.0075 1-propanol and P/P0 = 0.0083 2-propanol, where P and P0 are the partial pressure and standard vapor pressure, respectively, of a given analyte. The second task consisted of classification of a mixture of P/P0 = 0.011 1-propanol and P/P0 = 0.0090 2-propanol versus a mixture of P/P0 = 0.0090 1-propanol and P/P0 = 0.011 2-propanol. A third task consisted of multiple concentrations of three hydrocarbons, and a fourth task involved clustering two hydrocarbons in the presence of a variable background composition. An additional dataset was generated by exposing an array of five thin-film metal-oxide sensors to the headspace of seven different coffee blends. In each case, the MLP and FLD techniques were compared using the 5-sensor subset of the 20 available sensors that proved optimal for that dataset. The FLD and MLP algorithms yielded comparable performance on straightforward classification tasks, whereas the MLP technique yielded better performance on tasks that involved non-linear classification boundaries. In addition, for the four datasets produced by the carbon black-polymer composite detector array, the performance of each possible 5-sensor subset was evaluated using both signal processing approaches. The performance of the best 5-sensor subset selected with MLP was found to be slightly better than the performance of the FLD-selected subsets, and the performance of the median 5-sensor subset using MLP was nearer to that of the optimal subset than the median sensor array selected by FLD. In one case, the optimal test set performance distribution was found to be significantly better with MLP than with FLD: MLP had a clear advantage (86{\%} versus 57{\%} correct classification rate) when applied to the {"}coffees{"} dataset, and this trend is likely applicable to other multi-cluster classification tasks that consisted of non-Gaussian shaped data in lower-dimensional spaces.",
keywords = "Artificial neural networks, Fisher's linear discriminant, Multilayer perceptrons, Vapor sensors",
author = "Matteo Pardo and Sisk, {Brian C.} and Giorgio Sberveglieri and Lewis, {Nathan S}",
year = "2006",
month = "6",
day = "26",
doi = "10.1016/j.snb.2005.10.033",
language = "English",
volume = "115",
pages = "647--655",
journal = "Sensors and Actuators, B: Chemical",
issn = "0925-4005",
publisher = "Elsevier",
number = "2",

}

TY - JOUR

T1 - Comparison of Fisher's linear discriminant to multilayer perceptron networks in the classification of vapors using sensor array data

AU - Pardo, Matteo

AU - Sisk, Brian C.

AU - Sberveglieri, Giorgio

AU - Lewis, Nathan S

PY - 2006/6/26

Y1 - 2006/6/26

N2 - Two different classification methods, Fisher's linear discriminant (FLD) and a multilayer perceptron neural network (MLP), were directly compared with respect to their abilities to differentiate response patterns arising from arrays of chemical vapor detectors. The algorithms were compared in five different types of tasks that had been selected because they produced classification problems of varying character and difficulty. In one task, an array of 20 compositionally distinct carbon black-polymer composite vapor detectors was exposed to P/P0 = 0.0075 1-propanol and P/P0 = 0.0083 2-propanol, where P and P0 are the partial pressure and standard vapor pressure, respectively, of a given analyte. The second task consisted of classification of a mixture of P/P0 = 0.011 1-propanol and P/P0 = 0.0090 2-propanol versus a mixture of P/P0 = 0.0090 1-propanol and P/P0 = 0.011 2-propanol. A third task consisted of multiple concentrations of three hydrocarbons, and a fourth task involved clustering two hydrocarbons in the presence of a variable background composition. An additional dataset was generated by exposing an array of five thin-film metal-oxide sensors to the headspace of seven different coffee blends. In each case, the MLP and FLD techniques were compared using the 5-sensor subset of the 20 available sensors that proved optimal for that dataset. The FLD and MLP algorithms yielded comparable performance on straightforward classification tasks, whereas the MLP technique yielded better performance on tasks that involved non-linear classification boundaries. In addition, for the four datasets produced by the carbon black-polymer composite detector array, the performance of each possible 5-sensor subset was evaluated using both signal processing approaches. The performance of the best 5-sensor subset selected with MLP was found to be slightly better than the performance of the FLD-selected subsets, and the performance of the median 5-sensor subset using MLP was nearer to that of the optimal subset than the median sensor array selected by FLD. In one case, the optimal test set performance distribution was found to be significantly better with MLP than with FLD: MLP had a clear advantage (86% versus 57% correct classification rate) when applied to the "coffees" dataset, and this trend is likely applicable to other multi-cluster classification tasks that consisted of non-Gaussian shaped data in lower-dimensional spaces.

AB - Two different classification methods, Fisher's linear discriminant (FLD) and a multilayer perceptron neural network (MLP), were directly compared with respect to their abilities to differentiate response patterns arising from arrays of chemical vapor detectors. The algorithms were compared in five different types of tasks that had been selected because they produced classification problems of varying character and difficulty. In one task, an array of 20 compositionally distinct carbon black-polymer composite vapor detectors was exposed to P/P0 = 0.0075 1-propanol and P/P0 = 0.0083 2-propanol, where P and P0 are the partial pressure and standard vapor pressure, respectively, of a given analyte. The second task consisted of classification of a mixture of P/P0 = 0.011 1-propanol and P/P0 = 0.0090 2-propanol versus a mixture of P/P0 = 0.0090 1-propanol and P/P0 = 0.011 2-propanol. A third task consisted of multiple concentrations of three hydrocarbons, and a fourth task involved clustering two hydrocarbons in the presence of a variable background composition. An additional dataset was generated by exposing an array of five thin-film metal-oxide sensors to the headspace of seven different coffee blends. In each case, the MLP and FLD techniques were compared using the 5-sensor subset of the 20 available sensors that proved optimal for that dataset. The FLD and MLP algorithms yielded comparable performance on straightforward classification tasks, whereas the MLP technique yielded better performance on tasks that involved non-linear classification boundaries. In addition, for the four datasets produced by the carbon black-polymer composite detector array, the performance of each possible 5-sensor subset was evaluated using both signal processing approaches. The performance of the best 5-sensor subset selected with MLP was found to be slightly better than the performance of the FLD-selected subsets, and the performance of the median 5-sensor subset using MLP was nearer to that of the optimal subset than the median sensor array selected by FLD. In one case, the optimal test set performance distribution was found to be significantly better with MLP than with FLD: MLP had a clear advantage (86% versus 57% correct classification rate) when applied to the "coffees" dataset, and this trend is likely applicable to other multi-cluster classification tasks that consisted of non-Gaussian shaped data in lower-dimensional spaces.

KW - Artificial neural networks

KW - Fisher's linear discriminant

KW - Multilayer perceptrons

KW - Vapor sensors

UR - http://www.scopus.com/inward/record.url?scp=33646228613&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33646228613&partnerID=8YFLogxK

U2 - 10.1016/j.snb.2005.10.033

DO - 10.1016/j.snb.2005.10.033

M3 - Article

VL - 115

SP - 647

EP - 655

JO - Sensors and Actuators, B: Chemical

JF - Sensors and Actuators, B: Chemical

SN - 0925-4005

IS - 2

ER -