Gas Recognition Under Sensor Drift by Using Deep Learning
Machine olfaction is an intelligent system that combines cross-sensitivity chemical sensor array and an effective pattern recognition algorithm for the detection, identification, or quantification of various odors. Data collected by the sensor array are m
- PDF / 466,425 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 76 Downloads / 197 Views
Abstract Machine olfaction is an intelligent system that combines cross-sensitivity chemical sensor array and an effective pattern recognition algorithm for the detection, identification, or quantification of various odors. Data collected by the sensor array are multivariate time-series signals with complex structure, and these signals become more difficult to analyze due to sensor drift. In this work, we focus on improving the classification performance under sensor drift by using the deep learning method, which is popular among these years. Compared with other methods, our method can effectively tackle sensor drift by automatically extracting features, thus not only removing the complexity of designing the hand-made features, but also making it pervasive for a variety of application in machine olfaction. Our experimental results show that the deep learning method can learn the features that are more robust to drift than the original input and achieves high classification accuracy. Keywords Sensor drift
Deep learning Machine olfaction
X. Hu Q. Liu (&) H. Cai F. Li Digital Media Technology Key Laboratory of Sichuan Province, The School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China e-mail: [email protected] X. Hu e-mail: [email protected] H. Cai e-mail: [email protected] F. Li e-mail: [email protected]
Z. Wen and T. Li (eds.), Practical Applications of Intelligent Systems, Advances in Intelligent Systems and Computing 279, DOI: 10.1007/978-3-642-54927-4_3, Springer-Verlag Berlin Heidelberg 2014
23
24
X. Hu et al.
1 Introduction The principle of Machine olfaction is: (i) sensor array produces electrical signal response by absorbing the odor molecules; (ii) various methods are used to process these signals; (iii) pattern recognition system analyzes the input data and make decisions, accomplishing scheduled tasks, for instance, gas recognition and concentration measurement [1, 2]. In practical applications, sensor signals tend to show a significant variation due to changes in the experimental operating environment, aging of the sensor materials, and poisoning effects [3]. Sensor drift will change the cluster distribution in the data space and degrade the classification accuracy gradually [3, 4]. In the field of signal processing, the most effective methods for drift compensation are periodic recalibrations [4], whose main idea is using a reference gas to correct drift direction for individual sensor or for the entire array [5–7]. However, these techniques are all based on the assumption that the drift model is linear, which has not been confirmed yet, and they mostly need a reference gas that is chemically stable and highly correlated with the target gas in terms of sensor behavior, which is undoubtedly harsh in practical applications [4, 7]. Recently, machine learning methods are also applied to cope with drift [3, 8, 9]. In Vergara et al. [3], authors used ensemble learning methods to adapt a classifier to drift. This en
Data Loading...
