Standardization of metal oxide sensor array using artificial neural networks through experimental design

Zhang, Lei; Tian, Fengchun; Peng, Xiongwei; Dang, Lijun; Li, Guorui; Liu, Shouqiong; Kadri, Chaibou

doi:10.1016/j.snb.2012.11.113

Cited by 20 publications

(10 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This observation was also found in [12]. The performance of PDS (neural) is not good, for the neural network algorithm in [16] has nonlinear hidden neurons and is easy to overfit the transfer samples, especially when the number of input variables is large. The triangular window adopted in WPDS brings effective prior information to the problem, which makes it outperform other methods.…”

Section: Standardizationmentioning

confidence: 66%

“…There are three typical ways of calibration transfer [9,13]: transforming the data from the slave device to match the master one; updating the prediction model of the master device according to the slave data; and transforming the predicted values of the slave data. In the field of e-nose, focuses have been paid on the first way [8,9,[14][15][16][17], since it is feasible in most situations and easy to implement. This kind of methods are also known as device standardization methods, which essentially deal with a regression problem.…”

Section: Introductionmentioning

confidence: 99%

“…Common categories include univariate direct standardization (UDS), direct standardization (DS), and piecewise direct standardization (PDS), which differ mainly in the number of input variables. Regression algorithms such as robust fitting [8,17], artificial neural network (ANN) [14,16], partial least squares (PLS) [15], ordinary least square (OLS), and principal component regression (PCR) [12] have been studied. Besides, in [18], standardization was performed on a subspace obtained by http://dx.doi.org/10.1016/j.snb.2015.05.060 0925-4005/© 2015 Elsevier B.V. All rights reserved.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Improving the transfer ability of prediction models for electronic noses

Yan

Zhang

2015

Sensors and Actuators B: Chemical

View full text Add to dashboard Cite

Section: Standardizationmentioning

confidence: 66%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Improving the transfer ability of prediction models for electronic noses

Yan

Zhang

2015

Sensors and Actuators B: Chemical

View full text Add to dashboard Cite

“…Consequently, there has been a resurgence of interest in developing measurement techniques for air quality monitoring. Our previous work has proved that E-noses are an effective way to classify indoor pollutant gases [15,16]. …”

Section: Introductionmentioning

confidence: 99%

A Novel Semi-Supervised Electronic Nose Learning Technique: M-Training

Jia

Huang

Duan

et al. 2016

Sensors

View full text Add to dashboard Cite

When an electronic nose (E-nose) is used to distinguish different kinds of gases, the label information of the target gas could be lost due to some fault of the operators or some other reason, although this is not expected. Another fact is that the cost of getting the labeled samples is usually higher than for unlabeled ones. In most cases, the classification accuracy of an E-nose trained using labeled samples is higher than that of the E-nose trained by unlabeled ones, so gases without label information should not be used to train an E-nose, however, this wastes resources and can even delay the progress of research. In this work a novel multi-class semi-supervised learning technique called M-training is proposed to train E-noses with both labeled and unlabeled samples. We employ M-training to train the E-nose which is used to distinguish three indoor pollutant gases (benzene, toluene and formaldehyde). Data processing results prove that the classification accuracy of E-nose trained by semi-supervised techniques (tri-training and M-training) is higher than that of an E-nose trained only with labeled samples, and the performance of M-training is better than that of tri-training because more base classifiers can be employed by M-training.

show abstract

“…The kind of sensor discreteness must cause the reduction of E-nose prediction accuracy and reproducibility. To solve the problem of sensor discreteness, specific correction methods have been studied in previous publications [23,24]. Comparatively, the GAT-RWLS method proposed in [23], for its simplification of algorithm, is easier to implement for calibration in real-time E-nose detection.…”

Section: Introductionmentioning

confidence: 99%