有關人類疾病的研究長久以來被人們所重視，由於斑馬魚基因與人類相近且可以被快速繁殖，斑馬魚在相關研究上扮演著重要的角色。斑馬魚經常被用來研究藥物與毒物對神經系統與行為的影響，斑馬魚已經被用來模擬過阿茲海默症、腦癌、癲癇、焦慮症以及肝臟疾病。然而目前研究斑馬魚的實驗經常需要仰賴肉眼或是昂貴商用軟體器材紀錄斑馬魚的運動軌跡，並做簡單的分析。因此進行斑馬魚實驗所需的時間、人力與金錢成本，使斑馬魚實驗變得複雜。

本研究提出了一套斑馬魚行為模式分類系統Zebrafish Behavior Classification (ZBC) System，該系統能以軌跡資料對斑馬魚的行為進行分類。根據斑馬魚的啃咬、追逐、展示和正常等四種行為模式，我們提出了七種軌跡特徵的計算方法。這七種軌跡特徵包括軌跡相似度、總移動距離、平均幀間移動距離、最大和最小幀間移動距離、移動向量加總、最大和最小幀間移動距離向量，以及移動方向同方向的時間佔比。本系統會在計算七種軌跡特徵前先運用三種濾波方式清理軌跡資料，而七種軌跡特徵將用來訓練斑馬魚行為自動分類模型。

本研究比較了支持向量機 (Support Vector Machine, SVM)、隨機森林 (Random Forest, RF) 和極限梯度提升 (eXtreme Gradient Boosting, XGBoost) 等三種模型在斑馬魚行為自動分類模型上的準確度表現，同時我們也比較了均值濾波、中值濾波與卡爾曼濾波等三種資料清理方法運用在模型上的準確度表現。實驗結果顯示出ZBC在分類高度、中度和低度攻擊性的行為上能達到76%的準確度。另外，我們的研究發現不同資料清理方法和模型的搭配會影響分類的準確度，隨機森林在分類效果整體上有最高的準確度，而資料清理方法則須依照情況選擇。

Research in human diseases has long been a critical study, and zebrafish (Danio rerio) play an essential role in the field because of their genetic similarity to humans and can reproduce rapidly. Zebrafish are often used in experiments for the effects of drugs on behaviors, e.g., Alzheimer, brain cancer, epilepsy, anxiety, and liver disease. However, current experiments on zebrafish usually rely on the human eye or expensive commercial software to record the movement trajectory to analyze zebrafish behavior. As a result, the cost of zebrafish experiments in terms of time, labor, and money to make zebrafish experiments complicated to proceed.

In this study, we proposed a Zebrafish Behavior Classification (ZBC) System to classify zebrafish behaviors. Seven trajectory features are founded based on four behavioral patterns of zebrafish, including biting, chasing, displaying, and normal behavior. The seven features are trajectory similarity, total moving distance, maximum and minimum moving distance per frame, average moving distance per frame, summation of moving vector, maximum and minimum angle of moving vector per frame, and same direction ratio for two trajectories. The proposed system employs three filters to clean up trajectory data before estimating seven trajectory features. The estimated seven features are used to train the Zebrafish Behavior Classification Model.

This study compares the performance of accuracy on the Zebrafish Behavior Classification Model with three machine-learning methods - Support Vector Machine (SVM), Random Forest (RF), and eXtreme Gradient Boosting (XGBoost). The performance of data processed by the three filters (i.e., Mean Filter, Median Filter, Kalman Filter) is also compared to show their results. The results show that the ZBC can achieve 76% accuracy in classifying zebrafish behaviors regarding the level of aggression behaviors - high, moderate, and low aggression. In addition, our study found that the combination of data-cleaning methods and machine-learning models would affect classification performance. In general, the random forest had the best classification results, and the choice of filters depended on the situation.

[1] K. Dooley and L. I. Zon, “Zebrafish: a model system for the study of human disease,” Current Opinion in Genetics & Development, vol. 10, no. 3, pp. 252–256, Jun. 2000, doi: 10.1016/S0959-437X(00)00074-5.
[2] S. Saleem and R. R. Kannan, “Zebrafish: an emerging real-time model system to study Alzheimer’s disease and neurospecific drug discovery,” Cell Death Discovery, vol. 4, no. 1, p. 45, Oct. 2018, doi: 10.1038/s41420-018-0109-7.
[3] A. M. Stewart, O. Braubach, J. Spitsbergen, R. Gerlai, and A. V. Kalueff, “Zebrafish models for translational neuroscience research: from tank to bedside,” Trends in Neurosciences, vol. 37, no. 5, pp. 264–278, May 2014, doi: 10.1016/j.tins.2014.02.011.
[4] W. Goessling and K. C. Sadler, “Zebrafish: An Important Tool for Liver Disease Research,” Gastroenterology, vol. 149, no. 6, pp. 1361–1377, Nov. 2015, doi: 10.1053/j.gastro.2015.08.034.
[5] C.-W. Fu et al., “Exposure to silver impairs learning and social behaviors in adult zebrafish,” Journal of Hazardous Materials, vol. 403, p. 124031, Feb. 2021, doi: 10.1016/j.jhazmat.2020.124031.
[6] T. Capriello, M. C. Grimaldi, R. Cofone, S. D’Aniello, and I. Ferrandino, “Effects of aluminium and cadmium on hatching and swimming ability in developing zebrafish,” Chemosphere, vol. 222, pp. 243–249, May 2019, doi: 10.1016/j.chemosphere.2019.01.140.
[7] L. P. J. J. Noldus, A. J. Spink, and R. A. J. Tegelenbosch, “EthoVision: A versatile video tracking system for automation of behavioral experiments,” Behavior Research Methods, Instruments, & Computers, vol. 33, no. 3, pp. 398–414, Aug. 2001, doi: 10.3758/BF03195394.
[8] R. J. Egan et al., “Understanding behavioral and physiological phenotypes of stress and anxiety in zebrafish,” Behavioural Brain Research, vol. 205, no. 1, pp. 38–44, Dec. 2009, doi: 10.1016/j.bbr.2009.06.022.
[9] N. Zabinyakov et al., “Characterization of the first knock-out aldh7a1 zebrafish model for pyridoxine-dependent epilepsy using CRISPR-Cas9 technology,” PLOS ONE, vol. 12, no. 10, p. e0186645, Oct. 2017, doi: 10.1371/journal.pone.0186645.
[10] M. de O. Barreiros, D. de O. Dantas, L. C. de O. Silva, S. Ribeiro, and A. K. Barros, “Zebrafish tracking using YOLOv2 and Kalman filter,” Scientific Reports, vol. 11, no. 1, p. 3219, Feb. 2021, doi: 10.1038/s41598-021-81997-9.
[11] S. Zhuang et al., “Trajectory Tracking of Zebrafish Larvae Group,” in Robotic Micromanipulation of Zebrafish Larva, S. Zhuang, G. Zhang, D. Lei, X. Yu, M. Tong, W. Lin, Y. Shi, and H. Gao, Eds., Cham: Springer Nature Switzerland, 2023, pp. 233–256. doi: 10.1007/978-3-031-33410-8_8.
[12] C. Wang, Z. Wu, Y. Chen, W. Zhang, W. Ke, and Z. Xiong, “Improving 3D Zebrafish Tracking with Multi-View Data Fusion and Global Association,” IEEE Sensors Journal, pp. 1–1, 2023, doi: 10.1109/JSEN.2023.3288729.
[13] G. Si, F. Zhou, Z. Zhang, and X. Zhang, “Tracking Multiple Zebrafish Larvae Using YOLOv5 and DeepSORT,” in 2022 8th International Conference on Automation, Robotics and Applications (ICARA), Feb. 2022, pp. 228–232. doi: 10.1109/ICARA55094.2022.9738556.
[14] Y. Li et al., “A novel open-source raspberry Pi-based behavioral testing in zebrafish,” PLOS ONE, vol. 17, no. 12, p. e0279550, Dec. 2022, doi: 10.1371/journal.pone.0279550.
[15] P. Yang, H. Takahashi, M. Murase, and M. Itoh, “Zebrafish behavior feature recognition using three-dimensional tracking and machine learning,” Scientific Reports, vol. 11, no. 1, p. 13492, Jun. 2021, doi: 10.1038/s41598-021-92854-0.
[16] O. Anas et al., “Detecting Abnormal Fish Behavior Using Motion Trajectories In Ubiquitous Environments,” Procedia Computer Science, vol. 175, pp. 141–148, Jan. 2020, doi: 10.1016/j.procs.2020.07.023.
[17] W. Xu, Z. Zhu, F. Ge, Z. Han, and J. Li, “Analysis of Behavior Trajectory Based on Deep Learning in Ammonia Environment for Fish,” Sensors, vol. 20, no. 16, 2020, doi: 10.3390/s20164425.
[18] G. Wang, A. Muhammad, C. Liu, L. Du, and D. Li, “Automatic Recognition of Fish Behavior with a Fusion of RGB and Optical Flow Data Based on Deep Learning,” Animals, vol. 11, no. 10, 2021, doi: 10.3390/ani11102774.
[19] L. Yang et al., “A dual attention network based on efficientNet-B2 for short-term fish school feeding behavior analysis in aquaculture,” Computers and Electronics in Agriculture, vol. 187, p. 106316, Aug. 2021, doi: 10.1016/j.compag.2021.106316.
[20] A. Pérez-Escudero, J. Vicente-Page, R. C. Hinz, S. Arganda, and G. G. de Polavieja, “idTracker: tracking individuals in a group by automatic identification of unmarked animals,” Nature Methods, vol. 11, no. 7, pp. 743–748, Jul. 2014, doi: 10.1038/nmeth.2994.
[21] V. H. Sridhar, D. G. Roche, and S. Gingins, “Tracktor: Image-based automated tracking of animal movement and behaviour,” Methods in Ecology and Evolution, vol. 10, no. 6, pp. 815–820, Jun. 2019, doi: 10.1111/2041-210X.13166.
[22] C.-J. Chang, “Motion Trajectory Analysis of Zebrafish Based on Image Processing,” M.S. thesis, Dept. of Comput. Sci. & Inf. Eng., Nat. Taiwan Normal Univ., 2021. [Online]. Available: https://hdl.handle.net/11296/8a4xm2
[23] J. Cervantes, F. Garcia-Lamont, L. Rodríguez-Mazahua, and A. Lopez, “A comprehensive survey on support vector machine classification: Applications, challenges and trends,” Neurocomputing, vol. 408, pp. 189–215, Sep. 2020, doi: 10.1016/j.neucom.2019.10.118.
[24] L. Breiman, “Random Forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, Oct. 2001, doi: 10.1023/A:1010933404324.
[25] T. Chen and C. Guestrin, “XGBoost: A Scalable Tree Boosting System,” Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery, San Francisco, California, USA, pp. 785–794, 2016. [Online]. Available: https://doi.org/10.1145/2939672.2939785.
[26] K. N. Zabegalov et al., “Understanding zebrafish aggressive behavior,” Behavioural Processes, vol. 158, pp. 200–210, Jan. 2019, doi: 10.1016/j.beproc.2018.11.010.
[27] A. V. Kalueff et al., “Towards a Comprehensive Catalog of Zebrafish Behavior 1.0 and Beyond,” Zebrafish, vol. 10, no. 1, pp. 70–86, Mar. 2013, doi: 10.1089/zeb.2012.0861.
[28] J. E. Franco-Restrepo, D. A. Forero, and R. A. Vargas, “A Review of Freely Available, Open-Source Software for the Automated Analysis of the Behavior of Adult Zebrafish,” Zebrafish, vol. 16, no. 3, pp. 223–232, Jun. 2019, doi: 10.1089/zeb.2018.1662.
[29] N. Mehrasa, Y. Zhong, F. Tung, L. Bornn, G. Mori, and S. Fraser, “Deep Learning of Player Trajectory Representations for Team Activity Analysis,” in MIT Sloan Sports Analytics Conf., 2018.
[30] T. Maekawa et al., “Deep learning-assisted comparative analysis of animal trajectories with DeepHL,” Nature Communications, vol. 11, no. 1, p. 5316, Oct. 2020, doi: 10.1038/s41467-020-19105-0.
[31] I. Kontopoulos, A. Makris, and K. Tserpes, “A Deep Learning Streaming Methodology for Trajectory Classification,” ISPRS International Journal of Geo-Information, vol. 10, no. 4, 2021, doi: 10.3390/ijgi10040250.
[32] S. Cheng, K. Zhao, and D. Zhang, “Abnormal Water Quality Monitoring Based on Visual Sensing of Three-Dimensional Motion Behavior of Fish,” Symmetry, vol. 11, no. 9, 2019, doi: 10.3390/sym11091179.
[33] A. Teatini, “Movement trajectory classification using supervised machine learning,” Student thesis, 2019. Accessed: Dec. 10, 2019. [Online]. Available: http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-265009.
[34] W.-Y. Loh, “Classification and Regression Trees,” WIREs Data Mining and Knowl Discov, vol. 1, no. 1, pp. 14–23, Jan. 2011, doi: 10.1002/widm.8.
[35] M. Müller, Ed., “Dynamic Time Warping,” in Information Retrieval for Music and Motion, Berlin, Heidelberg: Springer Berlin Heidelberg, 2007, pp. 69–84. doi: 10.1007/978-3-540-74048-3_4.
[36] W.-C. Lee and J. Krumm, “Trajectory Preprocessing,” in Computing with Spatial Trajectories, Y. Zheng and X. Zhou, Eds., New York, NY: Springer New York, 2011, pp. 3–33. doi: 10.1007/978-1-4614-1629-6_1.
[37] Y. Zheng, “Trajectory Data Mining: An Overview,” ACM Trans. Intell. Syst. Technol., vol. 6, no. 3, May 2015, doi: 10.1145/2743025.
[38] K. Toohey and M. Duckham, “Trajectory Similarity Measures,” SIGSPATIAL Special, vol. 7, no. 1, pp. 43–50, May 2015, doi: 10.1145/2782759.2782767.
[39] J. Bergstra and Y. Bengio, “Random Search for Hyper-Parameter Optimization,” Journal of Machine Learning Research, vol. 13, no. 10, pp. 281–305, 2012.
[40] P. Banerjee, “A guide on XGBoost hyperparameters tuning,” Kaggle, https://www.kaggle.com/code/prashant111/a-guide-on-xgboost-hyperparameters-tuning (accessed Jul. 13, 2023).
[41] 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, Aug. 2020, doi: 10.1109/TKDE.2019.2912815.
[42] C.-C. Chang and C.-J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans. Intell. Syst. Technol., vol. 2, no. 3. Association for Computing Machinery, p. Article 27, 2011.
[43] L. E. Raileanu and K. Stoffel, “Theoretical Comparison between the Gini Index and Information Gain Criteria,” Annals of Mathematics and Artificial Intelligence, vol. 41, no. 1, pp. 77–93, May 2004, doi: 10.1023/B:AMAI.0000018580.96245.c6.