電子鼻(或稱仿生嗅覺系統)是一種透過氣體感測器來蒐集及辨識氣體樣本的儀器。然而，氣體感測器的反應常因生產時的個別差異或是感測器老化等問題影響產生感測器偏移。感測器偏移的現象會導致同一樣本會產生不一樣的反應，因而造成氣體辨識上的困難。
本研究透過選擇適當的氣體樣本以及遷移學習的方法，提出一套完整的流程來校正氣體感測器偏移的問題。為了減少校正時所須蒐集的氣體樣本數量，根據氣體感測器的特性，本研究提出由濃度來選擇氣體樣本的方法。透過選擇不同的氣體濃度，分類模型能夠更快的適應新的氣體資料的分布並在新的資料分布上保持原有的辨識率，降低感測器偏移所造成的辨識率下降等問題。
本研究將所提出的偏移校正方法應用在一個氣體感測器陣列資料集上，且實驗結果證明此方法是有效的。除此之外，整個學習的流程將會根據資料涵蓋時間的增加，使分類模型能夠被訓練成更為通用且不被感測器偏移所影響的分類模型。

The electronic nose (e-nose), a device that collects the responses of sensors to different gas samples and identifies their components, consists of multiple gas sensors and classification algorithms.

However, gas sensors suffer from problems related to both instrumental variation and aging, which result in different responses to even the same gas sample.
This alters the data distribution and reduces the accuracy of the classification.

This thesis proposes a transfer learning method called concentration based drift calibration (CDC) for calibrating the sensor drift as transfer samples are collected and then recalibrating the pre-trained model.
According to the characteristic of the metal-oxide gas sensors, the sensor response corresponds to the gas concentration.
That is, transfer samples are collected in the target domain at certain gas concentrations. After choosing the appropriate transfer samples, the pre-trained model adjusts their weights to fit the transfer samples and maintains the classification accuracy.

This method is evaluated on a complex time-varying drift dataset.
The experimental results show that the proposed method for drift calibration is effective, and can be used in real-world applications.
Moreover, the transfer process can be applied over time with data that has been previously collected to yield a more generalized model.

[1]J. W. Gardner and P. N. Bartlett, “A brief history of electronic noses,” Sensors and Actuators B: Chemical, vol. 18, no. 1-3, pp. 210–211, 1994.
[2] K. T. Tang, S. W. Chiu, C. H. Pan, H. Y. Hsieh, Y. S. Liang, and S. C. Liu, “Development of a portable electronic nose system for the detection and classification of fruity odors,” Sensors, vol. 10, no. 10, pp. 9179–9193, 2010.
[3] M. F. Adak and N. Yumusak, “Classification of e-nose aroma data of four fruit types by abc-based neural network,” Sensors, vol. 16, no. 3, p. 304, 2016.
[4] M. Peris and L. Escuder-Gilabert, “A 21st century technique for food control: Electronic noses,” Analytica chimica acta, vol. 638, no. 1, pp. 1–15, 2009.
[5] K. Brudzewski, S. Osowski, and A. Dwulit, “Recognition of coffee using differential electronic nose,” IEEE Transactions on Instrumentation and Measurement, vol. 61, no. 6, pp. 1803–1810, 2012.
[6] A. D’Amico, C. Di Natale, R. Paolesse, A. Macagnano, E. Martinelli, G. Pennazza,
M. Santonico, M. Bernabei, C. Roscioni, G. Galluccio et al., “Olfactory systems for medical applications,” Sensors and Actuators B: Chemical, vol. 130, no. 1, pp. 458–465, 2008.
[7] C. Di Natale, A. Macagnano, E. Martinelli, R. Paolesse, G. D’Arcangelo, C. Roscioni, A. Finazzi-Agro, and A. D’Amico, “Lung cancer identification by the analysis of breath by means of an array of non-selective gas sensors,” Biosensors and Bioelectronics, vol. 18, no. 10, pp. 1209–1218, 2003.
[8] E. J. Wolfrum, R. M. Meglen, D. Peterson, and J. Sluiter, “Metal oxide sensor arrays for the detection, differentiation, and quantification of volatile organic compounds at sub-parts-per-million concentration levels,” Sensors and Actuators B: Chemical, vol. 115, no. 1, pp. 322–329, 2006.
[9] A. D. Wilson and M. Baietto, “Applications and advances in electronic-nose technolo- gies,” Sensors, vol. 9, no. 7, pp. 5099–5148, 2009.
[10] A. Vergara, S. Vembu, T. Ayhan, M. A. Ryan, M. L. Homer, and R. Huerta, “Chemical gas sensor drift compensation using classifier ensembles,” Sensors and Actuators B: Chemical, vol. 166, pp. 320–329, 2012.
[11] S.-W. Chiu and K.-T. Tang, “Towards a chemiresistive sensor-integrated electronic nose: A review,” Sensors, vol. 13, no. 10, pp. 14 214–14 247, 2013.
[12] S. Guney and A. Atasoy, “Multiclass classification of n-butanol concentrations with k-nearest neighbor algorithm and support vector machine in an electronic nose,” Sensors and Actuators B: Chemical, vol. 166, pp. 721–725, 2012.
[13] R. Laref, E. Losson, A. Sava, and M. Siadat, “Support vector machine regression for calibration transfer between electronic noses dedicated to air pollution monitoring,” Sensors, vol. 18, no. 11, p. 3716, 2018.
[14]S. Feng, F. Farha, Q. Li, Y. Wan, Y. Xu, T. Zhang, and H. Ning, “Review on smart gas sensing technology,” Sensors, vol. 19, no. 17, p. 3760, 2019.
[15]L. Zhang and D. Zhang, “Efficient solutions for discreteness, drift, and disturbance (3d) in electronic olfaction,” IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 48, no. 2, pp. 242–254, 2016.
[16] L. Zhang, F. Tian, and D. Zhang, Electronic Nose: Algorithmic Challenges. Springer, 2018.
[17]W.-C. Wen, “An Enhanced Mixture Gas Prediction Algorithm by the Dynamic Re- sponse Analysis Based on the Temperature Modulation from a Metal Oxide Sensor,” Master’s thesis, National Tsing Hua University, Hsinchu, Taiwan, 2019.
[18] L. Zhang, F. Tian, L. Dang, G. Li, X. Peng, X. Yin, and S. Liu, “A novel background interferences elimination method in electronic nose using pattern recognition,” Sensors and Actuators A: Physical, vol. 201, pp. 254–263, 2013.
[19] R. O. Duda, P. E. Hart, and D. G. Stork, Pattern classification. John Wiley & Sons, 2012.
[20] S. J. Pan and Q. Yang, “A survey on transfer learning,” IEEE Transactions on knowledge and data engineering, vol. 22, no. 10, pp. 1345–1359, 2009.
[21] C. M. Bishop, Pattern recognition and machine learning. springer, 2006
[22] L. Zhang, F. Tian, S. Liu, L. Dang, X. Peng, and X. Yin, “Chaotic time series prediction of e-nose sensor drift in embedded phase space,” Sensors and Actuators B: Chemical, vol. 182, pp. 71–79, 2013.
[23] T. Artursson, T. Ekl¨ov, I. Lundstr¨om, P. M˚artensson, M. Sj¨ostr¨om, and M. Holmberg, “Drift correction for gas sensors using multivariate methods,” Journal of chemometrics, vol. 14, no. 5-6, pp. 711–723, 2000.
[24] M. Padilla, A. Perera, I. Montoliu, A. Chaudry, K. Persaud, and S. Marco, “Drift compensation of gas sensor array data by orthogonal signal correction,” Chemometrics and Intelligent Laboratory Systems, vol. 100, no. 1, pp. 28–35, 2010.
[25] K. Yan and D. Zhang, “Correcting instrumental variation and time-varying drift: a transfer learning approach with autoencoders,” IEEE Transactions on Instrumentation and Measurement, vol. 65, no. 9, pp. 2012–2022, 2016.
[26] L. Zhang, Y. Liu, Z. He, J. Liu, P. Deng, and X. Zhou, “Anti-drift in e-nose: A subspace projection approach with drift reduction,” Sensors and Actuators B: Chemical, vol. 253, pp. 407–417, 2017.
[27] L. Zhang and D. Zhang, “Domain adaptation extreme learning machines for drift com- pensation in e-nose systems,” IEEE Transactions on instrumentation and measurement, vol. 64, no. 7, pp. 1790–1801, 2014.
[28] K. Yan and D. Zhang, “Calibration transfer and drift compensation of e-noses via coupled task learning,” Sensors and Actuators B: Chemical, vol. 225, pp. 288–297, 2016.
[29] L. Zhang and D. Zhang, “Domain adaption extreme learning machines for drift compen- sation in e-nose systems,” IEEE Trans. Instrum. Meas., vol. 64, no. 7, pp. 1790–1801, Jul. 2015.
[30] Q. Liu, X. Li, M. Ye, S. S. Ge, and X. Du, “Drift compensation for electronic nose by semi-supervised domain adaption,” IEEE Sensors Journal, vol. 14, no. 3, pp. 657–665, 2013.