Academia.eduAcademia.edu

Figure 3 – uploaded by Vincenzo Naddeo

See full PDFdownloadDownload figure
Figure 3. Structure and topology of the artificial neural network pattern-recognition algorithms.

Figure 3 Structure and topology of the artificial neural network pattern-recognition algorithms.

Related Figures (16)

Table 1. Overview of the principal feature extraction methods and related features.
Table 1. Overview of the principal feature extraction methods and related features.
Table 2. Overview of the principal pattern-recognition techniques used in instrumental odour monitoring systems (IOMS). The research presents and discusses the influence of the application of different extracted signals and pattern recognition methods in the elaboration of the environmental odour classification monitoring model (OCMM) with IOMS. The paper aims to optimize the performance and robustness of an IOMS. The piecemeal signals (i.e., rise, intermediate, and peak state) obtained from the original response curves in combination with the use of the Linear Discriminant Analysis (LDA) or the Artificial Neural Networks (ANN) as pattern recognition techniques are investigated and argued. Laboratory experimental analysis with real samples were considered, to analyze and compare the results.
Table 2. Overview of the principal pattern-recognition techniques used in instrumental odour monitoring systems (IOMS). The research presents and discusses the influence of the application of different extracted signals and pattern recognition methods in the elaboration of the environmental odour classification monitoring model (OCMM) with IOMS. The paper aims to optimize the performance and robustness of an IOMS. The piecemeal signals (i.e., rise, intermediate, and peak state) obtained from the original response curves in combination with the use of the Linear Discriminant Analysis (LDA) or the Artificial Neural Networks (ANN) as pattern recognition techniques are investigated and argued. Laboratory experimental analysis with real samples were considered, to analyze and compare the results.
Table 3. Array of metal-oxide semiconductor gas sensors (MOS) present in the IOMS (seedOA). 2.3. Odour Classification Monitoring Model (OCMM) Elaboration 2.3.1. Data Acquisition
Table 3. Array of metal-oxide semiconductor gas sensors (MOS) present in the IOMS (seedOA). 2.3. Odour Classification Monitoring Model (OCMM) Elaboration 2.3.1. Data Acquisition
Figure 1. Electrical signals trend with respect to time at odour and odourless air exposure.
Figure 1. Electrical signals trend with respect to time at odour and odourless air exposure.
Figure 2. Extracted signals at different points (a) complete sensors response curve; (b) rise period; (c intermediate period; (d) peak period. The highest score indicates the group where those values belong. Meanwhile, artificial neural network (ANN) is biological paradigm that serves as mathematical models in simulating complex systems and considered black-box [47-52]. A general ANN consists of input neurons, hidden neurons, and output neurons, connected via synapse, which contains specific weight values [49,50]. For the experimental activities, a 3-layer feed-forward neural network was designed. The 13 different electrical resistance profiles from seedOA IOMS were used as input data, while the three investigated odour classes were used as target output (Figure 3). The ideal number of neurons is identified by means of “trial-and-error” on the basis of high correlation values (R*) and classification rates (%) between measured and predicted output.
Figure 2. Extracted signals at different points (a) complete sensors response curve; (b) rise period; (c intermediate period; (d) peak period. The highest score indicates the group where those values belong. Meanwhile, artificial neural network (ANN) is biological paradigm that serves as mathematical models in simulating complex systems and considered black-box [47-52]. A general ANN consists of input neurons, hidden neurons, and output neurons, connected via synapse, which contains specific weight values [49,50]. For the experimental activities, a 3-layer feed-forward neural network was designed. The 13 different electrical resistance profiles from seedOA IOMS were used as input data, while the three investigated odour classes were used as target output (Figure 3). The ideal number of neurons is identified by means of “trial-and-error” on the basis of high correlation values (R*) and classification rates (%) between measured and predicted output.
Table 4. Size of the training dataset for each of the 13 measurement sensors, at different extracted signals, by using linear discriminant analysis (LDA). For the training datasets, by applying the ANN, supervised learning was adopted. Binary classifiers, such as “1” and “0”, were assigned in the output to group the data, where “1” refers to the belonging to the group, while “0” indicates no interaction (Table 5).
Table 4. Size of the training dataset for each of the 13 measurement sensors, at different extracted signals, by using linear discriminant analysis (LDA). For the training datasets, by applying the ANN, supervised learning was adopted. Binary classifiers, such as “1” and “0”, were assigned in the output to group the data, where “1” refers to the belonging to the group, while “0” indicates no interaction (Table 5).
Table 5. Size of the training dataset for each of the 13 measurement sensors, at different extracted signals, by using an Artificial Neural Network (ANN).
Table 5. Size of the training dataset for each of the 13 measurement sensors, at different extracted signals, by using an Artificial Neural Network (ANN).
Table 6. Size of the validation dataset for each of the 13 measurement sensors, at different extracted signals. 2.4. Statistical Analysis Table 6. Size of the validation dataset for each of the 13 measurement sensors, at different extracted To assess the reliability of the trained models, a validation test was conducted, using the overall acquired data, the data excluded from the training dataset, and considering known the source for the comparison test (Table 6). The accuracy rates are, therefore, defined by analyzing the known values with the predicted ones.
Table 6. Size of the validation dataset for each of the 13 measurement sensors, at different extracted signals. 2.4. Statistical Analysis Table 6. Size of the validation dataset for each of the 13 measurement sensors, at different extracted To assess the reliability of the trained models, a validation test was conducted, using the overall acquired data, the data excluded from the training dataset, and considering known the source for the comparison test (Table 6). The accuracy rates are, therefore, defined by analyzing the known values with the predicted ones.
Table 7. Matrix of the different elaborated and compared odour classification monitoring models (OCMMs).
Table 7. Matrix of the different elaborated and compared odour classification monitoring models (OCMMs).
Table 8. Classification accuracy rate results with LDA application and training dataset. Table 8 summarizes the results of the classification accuracy rate obtained by applying the LDA model and the training dataset. Each column shows the classification rate per investigated piecemeal signal. The values of the Wilks’ lambda are also highlighted to analyze the degree of the discriminatory power of the model.
Table 8. Classification accuracy rate results with LDA application and training dataset. Table 8 summarizes the results of the classification accuracy rate obtained by applying the LDA model and the training dataset. Each column shows the classification rate per investigated piecemeal signal. The values of the Wilks’ lambda are also highlighted to analyze the degree of the discriminatory power of the model.
Table 9. Classification accuracy rate results with LDA application and validation dataset. Table 9 summarizes the classification metrics during the LDA validation test, deter. nined by using the discriminant factors equation developed in the training phase.
Table 9. Classification accuracy rate results with LDA application and validation dataset. Table 9 summarizes the classification metrics during the LDA validation test, deter. nined by using the discriminant factors equation developed in the training phase.
Figure 4. Scatterplots for the first two roots at different data extraction ((a) complete response curve; (b) rise period; (c) intermediate period; (d) peak period).
Figure 4. Scatterplots for the first two roots at different data extraction ((a) complete response curve; (b) rise period; (c) intermediate period; (d) peak period).
Table 10. Correlation values with ANN application during training phase.
Table 10. Correlation values with ANN application during training phase.
Figure 5. ANN performance at investigated extracted signal ((a) complete response curve; (b) rise period; (c) intermediate period; (d) peak period) with respects to mean square error versus the number of epochs.
Figure 5. ANN performance at investigated extracted signal ((a) complete response curve; (b) rise period; (c) intermediate period; (d) peak period) with respects to mean square error versus the number of epochs.
Table 11. Classification accuracy rate results with ANN application and validation dataset. The results show that the ANN misclassified Class C (ambient air) data. However, a perfect classification (100%) was achieved for Class A and Class B using the intermediate and peak data points. This scenario might be attributed to the idea that molecules of Class A and Class B are more sensitive to the gas sensors, in which an observable reaction
Table 11. Classification accuracy rate results with ANN application and validation dataset. The results show that the ANN misclassified Class C (ambient air) data. However, a perfect classification (100%) was achieved for Class A and Class B using the intermediate and peak data points. This scenario might be attributed to the idea that molecules of Class A and Class B are more sensitive to the gas sensors, in which an observable reaction
Table 12. Classification accuracy rates of all investigated conditions. The peak steady part is confirmed as the piecemeal signal that provides the highest discriminatory value for all the investigated cases and contains the most useful informa- tion for both the pattern-recognition techniques. Despite the complete response signals containing the complete information, this condition appears to slow down the performance of the pattern-recognition algorithm. A good classification accuracy was highlighted by using the intermediate period of the overall acquired data.
Table 12. Classification accuracy rates of all investigated conditions. The peak steady part is confirmed as the piecemeal signal that provides the highest discriminatory value for all the investigated cases and contains the most useful informa- tion for both the pattern-recognition techniques. Despite the complete response signals containing the complete information, this condition appears to slow down the performance of the pattern-recognition algorithm. A good classification accuracy was highlighted by using the intermediate period of the overall acquired data.
Related topics:
Computer ScienceArtificial IntelligenceAnalytical ChemistryMedicineLinear Discriminant AnalysisElectrical and Electronic Engineering
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved