本論文使用了基於16通道石英晶體微量天平電子鼻並結合暫態分析演算法，成功識別了5種肺炎致病菌在體外培養時不同接種濃度以及混合致病菌的有機揮發物。相對於過去的文獻，較少涉及使用嗅覺傳感器檢測細菌產生的揮發性有機化合物以識別體外培養環境中的菌液接種濃度或混合細菌。本論文的特色為透過電子鼻收集數據時的暫態響應作為主要分析訊號，配合從暫態響應萃取的特徵搭配機械學習演算法進行識別。以及探討過往文獻，分析每隻細菌容易產生的揮發性有機化合物與電子鼻頻道的關聯性。通過演算法中的特徵萃取，可以間接確定電子鼻通道的相對重要性。本論文中5種肺炎致病菌都以6種濃度（102/103/104/105/106 /107 CFU/ml）接種至培養基進行培養，本實驗對細菌濃度的分類任務中也加進了標準培養基樣品進行了分類。在混種細菌實驗下，分別進行3種細菌混菌（8類）與4種細菌混菌（15類）實驗。實驗中所有細菌都在培養基中培養24小時後才進行數據收集。在數據預處理中，採取了兩種步驟對原始暫態信號進行處理，即去除基線和數據歸一化。特徵生成階段從每個暫態信號中提取了斜率、積分、最大值、最大差異值、一階和二階微分等六種特徵。接續使用6種不同特徵萃取與6種機械學習辨識器來建構細菌濃度與混種致病菌的分類模型。最後使用留一量測交叉驗證和五折交叉驗證來評估最終的分類性能。這項研究利用電子鼻來檢測不同細菌產生的揮發性化合物。在25個萃取的特徵中至少超過一半以上為該隻細菌的主要揮發物通道。大腸桿菌釋放了吲哚、醛類和醇類等揮發性化合物，而電子鼻的通道2在其檢測中顯得至關重要，相關比例達76%。對於肺炎克雷伯菌，通道2表現出更高的重要性，可能與氨的排放有關。銅綠假單胞菌和金黃色葡萄球菌的化合物排放相似，但通道5對銅綠假單胞菌來說更為關鍵，這與氨、乙酸、醇和硫化氫的釋放量更高有關。無乳鏈球菌表現出獨特的揮發性化合物特徵，通道1和通道5對其有顯著的反應可用作區分無乳鏈球菌的關鍵生物標誌物的可能性。根據留一量測交叉驗證結果中，基於線性識別分析特徵萃取結合支持向量機分類器的分類方法識別細菌濃度的平均準確度達到99.28%。使用隨機森林分類器對所有細菌濃度（31類）進行辨識，分類準確率達到99.35%。在三種混合細菌的情況下，線性識別分析特徵萃取和支持向量機分類器的結合達到了99.67%的準確率。即使混種細菌數增加到4種，在同樣特徵萃取搭配K近鄰分類器準確率仍能保99.67%。細菌濃度分類結果不僅比以往的研究增加至少以往分類細菌濃度文獻中多數只分析3種細菌接種濃度類別，而此研究在辨識接種濃度提升到6種。而且相比與現有文獻最多的數據量，此研究收集了最大的測量數據集也提升至少3.6倍。在現有混種細菌文獻，最多只考量3種細菌混種的情況，分類總數共7類。分別為三隻菌單獨培養(C_1^3=3)、兩兩混種培養(C_2^3=3)、三隻菌混合培養(C_3^3=1)。在此研究將混種數目突破到4種，意味著四隻菌單獨培養(C_1^4=4)、兩兩混種培養(C_2^4=6)、三隻菌混合培養(C_3^4=4)與四隻菌(C_4^4=1)混合培養，共15種類別。這是前所未有的混合數量，即使需要識別更多的細菌，最終仍然保持高度識別準確率。這些結果也表明該暫態分析演算法對於細菌濃度與混種細菌的具備高度有效和可靠性。

This study utilized a 16-channel quartz crystal microbalance electronic nose combined with transient analysis algorithms to successfully identify the volatile organic compounds of five pneumonia-causing bacteria at different inoculation concentrations and mixed pathogenic bacteria in vitro. In contrast to previous literature, which lacked extensive investigation of using odor sensors to detect the volatile organic compounds produced by bacteria for identifying bacterial inoculation concentrations or mixed bacteria in vitro. The distinctive feature of this paper is the utilization of transient response as the main analytical signal for data collected by the electronic nose. It involves extracting features from the transient response and applying machine learning algorithms for recognition. n addition, the study delves into the literature to explore the association between the volatile organic compounds (VOCs) produced by each bacterium and the channels of the electronic nose. By employing feature extraction in the algorithm, the relative significance of the electronic nose channels can be indirectly ascertained. All five pneumonia-causing bacteria were cultured in six concentrations (102/103/104/105/106/107 CFU/ml) in the growth medium. Standard growth medium samples were also included in the bacterial concentration classification task. In the mixed-bacteria experiment, three different bacteria combinations (8 classes) and four different bacteria combinations (15 classes) were conducted. Data collection was conducted after 24 hours of incubation in the growth medium for all bacteria in the experiment. In the data preprocessing stage, two steps were employed to process the raw transient signals: baseline removal and data normalization. In the feature generation phase, six features were extracted from each transient signal, including slope, integration, maximum value, maximum difference value, first-order derivative, and second-order derivative. Six different feature extraction methods and six machine learning classifiers were used to construct classification models for bacterial concentrations and mixed pathogenic bacteria. The final performance of the models was evaluated using leave-one-measurement-out cross-validation and 5-fold cross-validation. This study employs an electronic nose to detect the volatile compounds produced by different bacteria. More than half of the 25 extracted features are identified as the primary volatile pathways for each bacterium. Escherichia coli releases volatile compounds like indole, aldehydes, and alcohols, with electronic nose channel 2 playing a crucial role in its detection, showing a correlation ratio of 76%. For Klebsiella pneumoniae, channel 2 exhibits higher significance, possibly linked to ammonia emissions. Pseudomonas aeruginosa and Staphylococcus aureus show similar compound emissions, but channel 5 is more critical for Pseudomonas aeruginosa, likely due to higher levels of ammonia, acetic acid, alcohols, and hydrogen sulfide. Streptococcus agalactiae exhibits unique volatile compound characteristics, with channels 1 and 5 showing significant responses, offering the potential for using them as key biological markers to differentiate Streptococcus agalactiae. In the leave-one-measurement-out cross-validation results, the classification method based on linear discriminant analysis feature extraction combined with support vector machine classifier achieved an average accuracy of 99.28% for identifying bacterial concentrations. The random forest classifier achieved a classification accuracy of 99.35% for recognizing all bacterial concentrations (31 classes). In the case of three mixed bacteria scenarios, the combination of linear discriminant analysis feature extraction and support vector machine classifier achieved an accuracy of 99.67%. Even when the number of mixed bacteria increased to four, the accuracy remained at 99.67% using the same feature extraction combined with KNN classifier. The classification results for bacterial concentrations not only surpass the majority of previous studies that typically analyzed only three bacterial inoculation concentration categories, but this study expands the recognition of inoculation concentrations to six. Moreover, compared to the largest existing datasets in similar research, this study collected the largest measurement dataset, which was at least 3.6 times larger. In previous studies on mixed bacteria, only three mixed bacteria scenarios were considered, resulting in a total of seven classes. These included individual cultures of three bacteria (C_1^3=3), pairwise mixed cultures of two bacteria (C_2^3=3), and mixed cultures of all three bacteria (C_3^3=1). In this study, the number of mixed bacteria scenarios was expanded to four, resulting in a total of 15 categories, including individual cultures of four bacteria (C_1^4=4), pairwise mixed cultures of two bacteria (C_2^4=6), mixed cultures of three bacteria (C_3^4=4), and mixed cultures of all four bacteria (C_4^4=1). This represents an unprecedented number of mixed scenarios, and even when identifying a larger number of bacterial species, the high recognition accuracy was maintained. These results demonstrate the high effectiveness and reliability of the transient analysis algorithm for bacterial concentration and mixed bacteria recognition.

[1] 衛生福利部統計處, “109年國人死因統計結果,” [Online serial]. Available: https://www.mohw.gov.tw/cp-5017-61533-1.html [Accessed Jun. 18, 2021].
[2] C. Cillóniz, R. Civljak, A. Nicolini, and A. Torres, “Polymicrobial community-acquired pneumonia: An emerging entity,” Respirology, vol. 21, no. 1, pp. 65–75, 2015.
[3] 台灣肺炎診治指引, “第三章：院內型肺炎3-1小節：常見菌種及流行病學,” [Online serial]. Available: https://pneumonia.idtaiwanguideline.org/guide/ch3-1.html#fn2 [Accessed May 8, 2022].
[4] M. H. Kollef, A. Shorr, Y. P. Tabak, V. Gupta, L. Z. Liu, and R. S. Johannes, “Epidemiology and outcomes of health-care-associated pneumonia: results from a large US database of culture-positive pneumonia,” Chest, vol. 128, no. 6, pp. 3854–3862, 2005.
[5] R. J. White, A. D. Blainey, K. J. Harrison, and S. K. Clarke, “Causes of pneumonia presenting to a district general hospital,” Thorax, vol. 36, no. 8, pp. 566–570, 1981.
[6] B. A. S. Pimentel, C. A. S. Martins, and J. C. Mendonça et al., “Streptococcus agalactiae infection in cancer patients: a five-year study,” European journal of clinical microbiology & infectious diseases, vol. 35, no. 6, pp. 927–933, 2016.
[7] R. Chaiwarith, W. Jullaket, and M. Bunchoo et al., “Streptococcus agalactiae in adults at chiang mai university hospital: a retrospective study,” BMC infectious diseases, vol. 11, no. 1, pp. 1471–2334, 2011.
[8] E. Scallan, R. M. Hoekstra, F. J. Angulo, R. V. Tauxe, M. A. Widdowson, S. L. Roy, J. L. Jones, and P. M. Griffin, “Foodborne illness acquired in the united states—major pathogens,” Emerging infectious diseases, vol. 17, no. 1, pp. 7–15, 2011.
[9] R. Chanderraj and R. P. Dickson, “Rethinking pneumonia: a paradigm shift with practical utility,” Proceedings of the national academy of sciences of the united states of america, vol. 115, no. 52, pp. 13148–13150, 2018.
[10] 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.
[11] K. Persaud and G. Dodd, “Analysis of discrimination mechanisms in the mammalian olfactory system using a model nose,” Nature, vol. 299, no. 5881, pp. 352–355, 1982.
[12] K. C. Persaud, “Electronic gas and odour detectors that mimic chemoreception in animals,” TrAC trends in analytical chemistry, vol. 11, no. 2, pp. 61–67, 1992.
[13] A. D. Wilson and M. Baietto, “Advances in electronic-nose technologies developed for biomedical applications,” Sensors, vol. 11, no. 1, pp. 1105–1176, 2011.
[14] H. Bai and G. Shi, “Gas sensors based on conducting polymers,” Sensors, vol. 7, no. 3, pp. 267–307, 2007.
[15] C. Wang, L. Yin, L. Zhang, D. Xiang, and R. Gao, “Metal oxide gas sensors: sensitivity and influencing factors,” Sensors, vol. 10, no. 3, pp. 2088–2106, 2010.
[16] A. Afzal, N. Iqbal, A. Mujahid, and R. Schirhagl, “Advanced vapor recognition naterials for selective and fast responsive surface acoustic wave sensors: a review,” Analytica chimica acta, vol. 787, pp. 36–49, 2013.
[17] X. Liu, S. Cheng, H. Liu, S. Hu, D. Zhang, and H. Ning, “A survey on gas sensing technology,” Sensors, vol. 12, no. 7, pp. 9635–9665, 2012.
[18] B. Wyszynski and T. Nakamoto, “Linking biological and artificial olfaction: biomimetic quartz crystal microbalance odor sensors,” IEEJ transactions on electrical and electronic engineering, vol. 4, no. 3, pp. 334–338, 2009.
[19] A. D. Alphus Dan Wilson, “Recent progress in the design and clinical development of electronic-nose technologies,” Nanobiosensors in disease diagnosis, p. 15, 2016.
[20] Y. Wang, A. Liu, Y. Han, and T. Li, “Sensors based on conductive polymers and their composites: a review,” Polymer international, vol. 69, no. 1, pp. 7–17, 2019.
[21] H. Ji, W. Zeng, and Y. Li, “Gas sensing mechanisms of metal oxide semiconductors: a focus review,” Nanoscale, vol. 11, no. 47, pp. 22664–22684, 2019.
[22] J. R. Askim, M. Mahmoudi, and K. S. Suslick, “Optical sensor arrays for chemical sensing: the optoelectronic nose,” Chemical society reviews, vol. 42, no. 22, p. 8649, 2013.
[23] X. Xu, H. Cang, C. Li, Z. K. Zhao, and H. Li, “Quartz crystal microbalance sensor array for the detection of volatile organic compounds,” Talanta, vol. 78, no. 3, pp. 711–716, 2009.
[24] T. M. A. Gronewold, “Surface acoustic wave sensors in the bioanalytical field: recent trends and challenges,” Analytica chimica acta, vol. 603, no. 2, pp. 119–128, 2007.
[25] F. Lough, “Detection of exogenous vocs as a novel in vitro diagnostic technique for the detection of pathogenic bacteria,” Trends in analytical chemistry, vol. 87, pp. 71–81, 2017.
[26] A. W. Boots, A. Smolinska, J. J. van Berkel, R. R. Fijten, E. E. Stobberingh, M. L. Boumans, E. J. Moonen, E. F. Wouters, J. W. Dallinga, and F. J. Van Schooten, “Identification of microorganisms based on headspace analysis of volatile organic compounds by gas chromatography-mass spectrometry,” Journal of breath research, vol. 8, no. 2, p. 027106, 2014.
[27] R. M. S. Thorn, D. M. Reynolds, and J. Greenman, “Multivariate analysis of bacterial volatile compound profiles for discrimination between selected species and strains in vitro,” Journal of microbiological methods, vol. 84, no. 2, pp. 258-264, 2011.
[28] T. Seesaard, C. Thippakorn, T. Kerdcharoen, and S. Kladsomboon, “A hybridelectronic nose system for discrimination of pathogenic bacterial volatile compounds,” Analytical methods, vol. 12, no. 47, pp. 5671–5683, 2020.
[29] F. Wojciech, Ż. Karolina, M. Marta, D. Dagmara, B. Tomasz, W. Natalia, and B. Barbara, “GC-MS profiling of volatile metabolites produced by Klebsiella pneumoniae,” Frontiers in molecular biosciences, vol. 9, 2022.
[30] O. Lawal, H. Muhamadali, W. M. Ahmed, I. R. White, T. M. E. Nijsen, R. Goodacre, and S. J. Fowler, “Headspace volatile organic compounds from bacteria implicated in ventilator-associated pneumonia analysed by TD-GC/MS,” Journal of breath research, vol. 12, no. 2, p. 026002, 2018.
[31] J. Zhu, H. D. Bean, Y. M. Kuo, and J. E. Hill, “Fast detection of volatile organic compounds from bacterial cultures by secondary electrospray ionization-mass spectrometry,” Journal of clinical microbiology, vol. 48, no. 12, pp. 4426-4431, 2010.
[32] Z. Wang, M. Y. Li, B. Peng, Z. X. Cheng, H. Li, and X. X. Peng, “GC–MS-based metabolome and metabolite regulation in serum-resistant Streptococcus agalactiae,” Journal of proteome research, vol. 15, no. 7, pp. 2246-2253, 2016.
[33] P. E. Fournier, M. Drancourt, P. Colson, J.-M. Rolain, B. L. Scola, and D. Raoult, “Modern clinical microbiology: new challenges and solutions,” Nature reviews microbiology, vol. 11, no. 8, pp. 574–585, 2013
[34] A.-L. Välimaa, A. Tilsala-Timisjärvi, and E. Virtanen, “Rapid detection and identification methods for listeria monocytogenes in the food chain – a review,” Food control, vol. 55, pp. 103–114, 2015.
[35] C. I. L. Justino, A. C. Duarte, and T. A. P. Rocha-Santos, “Recent progress in biosensors for environmental monitoring: a review,” Sensors, vol. 17, no. 12, p. 2918, 2017.
[36] P. Boilot, E.L. Hines, J.W. Gardner, R. Pitt, S. John, J. Mitchell, and D.W. Morgan, “Classification of bacteria responsible for ENT and eye infections using the Cyranose system,” IEEE sensors journal, vol. 2, no. 3, pp. 247–253, 2002.
[37] R. Dutta, A. Das, N. G. Stocks, and D. Morgan, “Stochastic resonance-based electronic nose: a novel way to classify bacteria,” Sensors and actuators B: chemical, vol. 115, no. 1, pp. 17–27, 2006.
[38] R. Fend, A. H. J. Kolk, C. Bessant, P. Buijtels, P. R. Klatser, and A. C. Woodman, “Prospects for clinical application of electronic-nose technology to early detection of Mycobacterium tuberculosis in culture and sputum,” Journal of clinical microbiology, vol. 44, no. 6, pp. 2039–2045, 2006.
[39] G. C. Green, A. D. C. Chan, and R. A. Goubran, “Dimensionality reduction methods of electronic nose data for bacteria discrimination,” CMBES proc., vol. 30, no. 1, Dec. 2007.
[40] P. F. Astantri, W. S. Prakoso, K. Triyana, T. Untari, C. M. Airin, and P. Astuti, “Lab-made electronic nose for fast detection of listeria monocytogenes and bacillus cereus,” Veterinary sciences, vol. 7, no. 1, p. 20, 2020.
[41] W. H. Tay, K. K. L. Chong, and K. A. Kline, “Polymicrobial–host interactions during infection,” Journal of molecular biology, vol. 428, no. 17, pp. 3355–3371, 2016.
[42] H. Trill and S. Setford, “Application of a single sensor odour analyser as a diagnostic tool for the detection and identification of pathogenic wound bacteria,” Cranfield university, 2007.
[43] N. Yusuf, A. Zakaria, M. I. Omar, A. Y. M. Shakaff, M. J. Masnan, L. M. Kamarudin, N. A. Rahim, N. Z. I. Zakaria, A. A. Abdullah, A. Othman, and M. S. Yasin, “In-vitro diagnosis of single and poly microbial species targeted for diabetic foot infection using E-nose technology,” BMC bioinformatics, vol. 16, no. 1, p. 158, 2015.
[44] E. Núñez Carmona, M. Soprani, and V. Sberveglieri, “Nanowire (S3) device for the quality control of drinking water,” Sensors for everyday life, vol. 23, pp. 179–203, 2016.
[45] H. Sun, F. Tian, Z. Liang, T. Sun, B. Yu, S. X. Yang, Q. He, L. Zhang, and X. Liu, “Sensor array optimization of electronic nose for detection of bacteria in wound infection,” IEEE transactions on industrial electronics, vol. 64, no. 9, pp. 7350–7358, 2017.
[46] Z. Liang, F. Tian, C. Zhang, H. Sun, A. Song, and T. Liu, “Improving the robustness of prediction model by transfer learning for interference suppression of electronic nose,” IEEE sensors journal, vol. 18, no. 3, pp. 1111–1121, 2018.
[47] J. R. Stetter and W. R. Penrose, “Understanding chemical sensors and chemical sensor arrays (electronic noses): past, present, and future,” Sensor applications, vol. 10, no. 1, pp. 189–229, 2002.
[48] J. W. Gardner, P. Boilot, and E. L. Hines, “Enhancing electronic nose performance by sensor selection using a new integer-based genetic algorithm approach,” Sensors and actuators B: chemical, vol. 106, no. 1, pp. 114–121, 2005.
[49] M. Egashira, Y. Shimizu, and Y. Takao, “Trimethylamine sensor based on semiconductive metal oxides for detection of fish freshness,” Sensors and actuators B: chemical, vol. 1, no. 1–6, pp. 108–112, 1990.
[50] T. C. Pearce, J. W. Gardner, S. Friel, P. N. Bartlett, and N. Blair, “Electronic nose for monitoring the flavour of beers,” The analyst, vol. 118, no. 4, p. 371, 1993.
[51] S. Zhang, C. Xie, D. Zeng, Q. Zhang, H. Li, and Z. Bi, “A feature extraction method and a sampling system for fast recognition of flammable liquids with a portable E-nose,” Sensors and actuators B: chemical, vol. 124, no. 2, pp. 437–443, 2007.
[52] N. Nimsuk, “Improvement of accuracy in beer classification using transient features for electronic nose technology,” Journal of food measurement and characterization, vol. 13, no. 1, pp. 656–662, 2018.
[53] C. Distante, M. Leo, P. Siciliano, and K. C. Persaud, “On the study of feature extraction methods for an electronic nose,” Sensors and actuators B: chemical, vol. 87, no. 2, pp. 274–288, 2002.
[54] T. Eklöv, P. Mårtensson, and I. Lundström, “Enhanced selectivity of MOSFET gas sensors by systematical analysis of transient parameters,” Analytica chimica acta, vol. 353, no. 2–3, pp. 291–300, 1997.
[55] A. K. M. S. Islam, Z. Ismail, M. N. Ahmad, B. Saad, A. R. Othman, A. Y. M. Shakaff, A. Daud, and Z. Ishak, “Transient parameters of a coated quartz crystal microbalance sensor for the detection of volatile organic compounds (vocs),” Sensors and actuators B: chemical, vol. 109, no. 2, pp. 238–243, 2005.
[56] L. Kou, D. Zhang, and D. Liu, “A novel medical E-nose signal analysis system,” Sensors, vol. 17, no. 4, p. 402, 2017.
[57] J. Huang and J. Wu, “Robust and rapid detection of mixed volatile organic compounds in flow through air by a low cost electronic nose,” Chemosensors, vol. 8, no. 3, p. 73, 2020.
[58] J. Kankare, “Sauerbrey equation of quartz crystal microbalance in liquid medium,” Langmuir, vol. 18, no. 18, pp. 7092–7094, 2002.
[59] B. I. Ismail, E. M. Goortani, M. B. A. Karim, W. M. Tat, S. Setapa, J. Y. Luke, and O. H. Hoe, “Evaluation of docker as edge computing platform,” 2015 IEEE conference on open systems (ICOS), 2015.
[60] D. Liu and L. Zhao, “The research and implementation of cloud computing platform based on Docker,” 2014 11th International computer conference on wavelet actiev media technology and information processing (ICCWAMTIP), 2014.
[61] Y. Y. Song and Y. Lu, “Decision tree methods: applications for classification and prediction,” Shanghai arch psychiatry, vol. 27, no. 2, pp. 130-135, 2015.
[62] M. A. Berry and G. S. Linoff, “Mastering data mining: the art and science of customer relationship management,” Industrial management & amp; data systems, vol. 100, no. 5, pp. 245–246, 2000.
[63] S. Santosh, G. Maya, and F. Andrew, “Bayesian quadratic discriminant analysis,” Journal of machine learning research, vol. 8, no. 6, pp. 1277-1305, 2007.
[64] S. Bose, A. Pal, R. Saharay, and J. Nayak, “Generalized quadratic discriminant analysis,” Pattern recognition, vol. 48, no. 8, pp. 2676-2684, 2015.
[65] C. Croux and K. Joossens, “Influence of observations on the misclassification probability in quadratic discriminant analysis,” Multivariate analysis, vol. 96, no. 2, pp. 384-403, 2005.
[66] D. T. Larose and C. D. Larose, “Discovering knowledge in data: an introduction to data mining,” John wiley & sons, vol. 4, 2014.
[67] L. Ma, M. M. Crawford, and J. Tian, “Local manifold learning-based k-nearest-neighbor for hyperspectral image classification,” IEEE transactions on geoscience and remote sensing, vol. 48, no. 11, pp. 4099-4109, 2010.
[68] S. Sun and R. Huang, “An adaptive k-nearest neighbor algorithm,” 2010 Seventh international conference on fuzzy systems and knowledge discovery, vol. 1, pp. 91-94, 2010.
[69] M. L. Zhang and Z. H. Zhou, “A k-nearest neighbor based algorithm for multi-label classification,” 2005 IEEE international conference on granular computing, beijing, vol. 2, pp. 718-721, 2005.
[70] N. García-Pedrajas, J. A. R. del Castillo, and G. Cerruela-García, “A proposal for local k values for k-nearest neighbor rule,” IEEE transactions on neural networks and learning systems, vol. 28, no. 2, pp. 470-475, 2017.
[71] S. A. Dudani, “The distance-weighted k-nearest-neighbor rule,” IEEE transactions on systems, man, and cybernetics, vol. SMC-6, no. 4, pp. 325-327, 1976.
[72] W. Y. Loh, “Classification and regression trees,” Wiley interdisciplinary reviews: data mining and knowledge discovery, vol. 1, no. 1, pp. 14-23, 2011.
[73] G. Biau and L. Devroye, “On the layered nearest neighbour estimate, the bagged nearest neighbour estimate and the random forest method in regression and classification,” Multivariate analysis, vol. 101, no. 10, pp. 2499-2518, 2010.
[74] X. Chen and H. Ishwaran, “Random forests for genomic data analysis,” Genomics, vol. 99, no. 6, pp. 323–329, 2012.
[75] T. Chen and C. Guestrin, “XGBoost: a scalable tree boosting system,” Association for computing machinery, pp. 785-794, 2016.
[76] A. Gupta, K. Gusain, B. Popli, “Verifying the value and veracity of extreme gradient boosted decision trees on a variety of datasets,” 2016 11th International conference on industrial and information systems (ICIIS), pp. 457-162, 2016.
[77] G. Wang, “A survey on training algorithms for support vector machine classifiers,” 2008 Fourth international conference on networked computing and advanced information management, vol. 1, pp. 123-128, 2008.
[78] A. Tharwat, “Parameter investigation of support vector machine classifier with kernel functions,” Knowledge and information systems, vol. 61, no. 3, pp. 1269-1302, 2019.
[79] S. Amari and S. Wu, “Improving support vector machine classifiers by modifying kernel functions,” Neural networks, vol. 12, no. 6, pp. 783-789, 1999.
[80] P. Refaeilzadeh, L. Tang, and H. Liu, “Cross-validation,” Encyclopedia of database systems, pp. 532–538, 2009.
[81] D. Berrar, “Cross-validation,” Encyclopedia of bioinformatics and computational biology, vol. 1, pp. 542-545, 2019.
[82] T. T. Wong and P. Y. Yeh, “Reliable accuracy estimates from k-fold cross validation,” IEEE transactions on knowledge and data engineering, vol. 32, no. 8, pp. 1586-1594, 2020.
[83] K. Damir, B. Maja, Š. Ljiljana, and B. Š. Dunja, “Multi-label classifier performance evaluation with confusion matrix,” Computer science & information technology, vol. 10, no. 08, p. 1, 2020.
[84] T. Robert, “Sensitivity, specificity, and predictive values: foundations, pliabilities, and pitfalls in research and practice,” Frontiers in public health, vol. 5, p. 307, 2017.
[85] S. Marina, J. Nathalie, and S. Stan, “Beyond accuracy, f-score and ROC: a family of discriminant measures for performance evaluation,” AI 2006: advances in artificial intelligence, lecture notes in computer science, vol. 4304, pp. 1015-1021, 2006.
[86] T. Fawcett, “An introduction to roc analysis,” Pattern recognition letter, vol. 27, no. 8, pp. 861–874, 2006.
[87] J. N. Mandrekar, “Receiver operating characteristic curve in diagnostic test assessment,” Thoracic oncology, vol. 5, no. 9, pp. 1315–1316, 2010.
[88] Y. C. Tung, “Development of an AI odor recognition edge computing system using QCM-based electronic nose and applying in pneumonia pathogens odors recognition under in vitro culture environment,” M. S. thesis, Dept. BME, National Cheng Kung University, 2022.