Odor Detection Using an E-Nose With a Reduced Sensor Array

Abstract

Recent advances in the field of electronic noses (e-noses) have led to new developments in both sensors and feature extraction as well as data processing techniques, providing an increased amount of information. Therefore, feature selection has become essential in the development of e-nose applications. Sophisticated computation techniques can be applied for solving the old problem of sensor number optimization and feature selections. In this way, one can find an optimal application-specific sensor array and reduce the potential cost associated with designing new e-nose devices. In this paper, we examine a procedure to extract and select modeling features for optimal e-nose performance. The usefulness of this approach is demonstrated in detail. We calculated the model’s performance using cross-validation with the standard leave-one-group-out and group shuffle validation methods. Our analysis of wine spoilage data from the sensor array shows when a transient sensor response is considered, both from gas adsorption and desorption phases, it is possible to obtain a reasonable level of odor detection even with data coming from a single sensor. This requires adequate extraction of modeling features and then selection of features used in the final model.

Application of a Low-Cost Electronic Nose for Differentiation between Pathogenic Oomycetes Pythium intermedium and Phytophthora plurivora

“…The further study that we performed using collected data is a verification of the possibility of further reducing the price of the low-cost electronic nose that we construct. As we demonstrated using different datasets [ 61 ], it may be feasible to reduce the sensor array and retain only a small portion of sensors, or even one sensor. Despite such reduction, the classification models’ performance, built on features extracted from the dynamic sensors response curves, allows achieving a more similar classification performance than the one obtained using the data collected by all sensors.…”

“…What is more, these features are calculated for the whole sensor response curve and the adsorption and desorption phases separately. The full list of the features that we have used for training classification models is presented in Appendix A in the other report [ 61 ].…”

Instrumental Odour Monitoring System Classification Performance Optimization by Analysis of Different Pattern-Recognition and Feature Extraction Techniques

“…The training dataset was used to determine the coefficients of the two investigated mathematical models, considered subsequently for the validation stage. The validation dataset, consisting of six separate sets of samples and applied according to “leave-one-group-out” method [ 24 ], was adopted as test samples to verify the model accuracies.…”

“…Many studies feed the complete signals as input to the pattern-recognition algorithm, which makes the system computationally expensive, complex, and hard to implement and requires a large memory space [ 19 , 22 ]. Due to large number of values, feature extraction of signals is used to eliminate redundant data and improve the accuracy of the IOMS [ 24 , 25 , 26 , 27 ]. By applying this method, the most important data from large set of signals can be captured, resulting in a reduction in computation time, as well as an increase in the speed of measurement and storage [ 28 , 29 ].…”

Collaborative Filtering to Predict Sensor Array Values in Large IoT Networks

“…There exists also proposals that make use of deep learning through recurrent neural networks, for improving the accuracy in the concentration identification of gas mixtures in gas sensor array raw data, as in [8]. Finally, feature selection can be a main stage in the sensor arrays data analysis: in [9] the authors make use of an e-nose applied to odor-detection and they report that this task requires adequate extraction and selection of features that will be used in the final model.…”