人工智慧（AI）已被廣泛應用於醫療保健、金融、零售和交通等各個領域。然而，AI最具挑戰性的問題是黑盒子問題。為了解決這個問題，可解釋性人工智慧（XAI）已成為學術界一個熱門且重要的議題。儘管過去已發明和應用了許多知名的XAI算法，仍然存在一些常見問題。
本研究結合因果推斷和迴歸模型，使用最具解釋能力和近似於人類思考模式的反事實方法來獲得解釋性結果。以下步驟用於獲得結果：1. 每個特徵生成適量的個別變化，2. 透過計算每個特徵及其相應的變化生成反事實數據，並使用訓練好的模型預測結果，3. 計算原始數據的預測結果與反事實數據的預測結果之間的差異，以及4. 以特徵變化為自變數、預測結果差異為反應變物以擬合三次樣條迴歸模型(Cubic Splines Regression Model)。
迴歸模型生成的迴歸線以及特徵變化與差異之間的關係揭示了每項特徵在數據中經歷固定變化時預測結果變化的程度和趨勢。這包括預測結果變大或變小、哪些特徵對預測結果有顯著影響，以及哪些特徵在影響預測結果時可能與其他特徵產生交互作用。該方法不僅同時實現了局部和全局解釋，而且與預測結果具有高度的忠實度(Fidelity)和一致性。此外解釋過程高度可解釋性，不過度依賴數學理論。

Artificial intelligence (AI) has been widely applied in various fields such as healthcare, finance, retail, and transportation. However, the black box problem is the most challenging issue in the field of AI. To solve this problem, explainable AI (XAI) has become a popular and important topic in the academic community. Despite the invention and application of many well-known XAI algorithms, there are still common issues.
This study uses the most interpretable and human-like counterfactual method in combination with causal inference and regression models to obtain explanatory results. The following steps are used to obtain the results: (1) each feature generates an appropriate amount of individual changes, (2) the counterfactual data is generated by calculating each feature and its corresponding change, and predicting the results using the trained model, (3) the difference between the predicted result of the original data and the predicted result of the counterfactual data is calculated, and (4) the regression model is fitted with feature change as the independent variable and the difference as the dependent variable.
The regression line generated by the regression model and the relationship between feature changes and the difference reveal the degree and trend of the predicted result changes when each feature undergoes a fixed change in the data. This method not only achieves both local and global explanations but also has high fidelity and consistency with the prediction results. Additionally, the explanatory process is highly interpretable without relying heavily on mathematical theories.

[1] Akaike, H. A new look at the statistical model identification. IEEE Transactions on Automatic Control, vol. 19, no. 6, pp. 716-723, doi: 10.1109/TAC.1974.1100705. (1974).
[2] Adadi, A., Berrada, M. Explainable AI for Healthcare: From Black Box to Interpretable Models. In: Bhateja, V., Satapathy, S., Satori, H. (eds) Embedded Systems and Artificial Intelligence. Advances in Intelligent Systems and Computing, vol 1076. Springer, Singapore. https://doi.org/10.1007/978-981-15-0947-6_31. (2020).
[3] Akhai, S. From Black Boxes to Transparent Machines: The Quest for Explainable AI. Available at SSRN: https://ssrn.com/abstract=4390887 or http://dx.doi.org/10.2139/ssrn.4390887. (2023).
[4] Breiman, L. Random Forests. Machine Learning 45, 5–32. https://doi.org/10.1023/A:1010933404324 (2001).
[5] Burnham, K. P. and Anderson, D. R. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Springer New York, NY. ISBN= 978-0-387-22456-5. (2002).
[6] Brughmans, D., Leyman, P. and Martens, D. NICE: An Algorithm for Nearest Instance Counterfactual Explanations. arXiv. https://doi.org/10.48550/arXiv.2104.07411. (2022).
[7] Chattopadhyay, A., Manupriya, P., Sarkar, A. and Balasubramanian, V. N. Neural Network Attributions: A Causal Perspective. arXiv. DOI= 10.48550/ARXIV.1902.02302. Available from https://arxiv.org/abs/1902.02302. (2019).
[8] Demajo, L. M., Vella, V. and Dingli, A. Explainable AI for Interpretable Credit Scoring. AIRCC Publishing Corporation. https://arxiv.org/abs/2012.03749 (2020).
[9] Dou, B., Zhu Z., Merkujev, E., Ke, L., Chen, L., Jiang, J., Zhu, Y., Liu J., Zhang, B. and Wei, G. W. Machine Learning Methods for Small Data Challenges in Molucular Science. ACS Publications. DOI=10.1021/acs.chemrev.3c00189. (2023).
[10] Efron, B., Hastie, T., Johnstone, T. and Tibshirani, R. “Least Angle Regression,” Annals of Statistics (with discussion), 407-499. (2004).
[11] Glass, T. A., Goodman, S. N., Hernan, M. A. and Samet J. M. Causal Inference in Public Health. Annu. Rev. Public Health 2013. 34:61–75. Doi: 10.1146/annurev-publhealth-031811-124606. (2013).
[12] Goyal, Y., Wu, Z., Ernst, J., Batra, D. Parikh, D. and Lee, S. Counterfactual Visual Explanations. arXiv:1904.07451. (2019).
[13] James, G., Witten, D., Hastie, T. and Tibshirani, R. An Introduction to Statistical Learning: with Applications in R. Springer New York, NY. DOI: https://doi.org/10.1007/978-1-0716-1418-1. (2021).
[14] Joyce, D. W., Kormilitzin, A., Smith, K. A. and Cipriani, A. Explainable artificial intelligence for mental health through transparency and interpretability for understandability. npj Digit. Med. 6, 6 (2023). https://doi.org/10.1038/s41746-023-00751-9. (2023).
[15] Karimi, A., Barthe, G., Balle, B. and Valera, I. Model-Agnostic Counterfactual Explanations for Consequential Decisions. Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics., in Proceedings of Machine Learning Research. 108:895-905 Available from https://proceedings.mlr.press/v108/karimi20a.html. (2020).
[16] Kamath, U. and Liu, J. Explainable Artificial Intelligence: An Introduction to Interpretable Machine Learning. Springer Cham. ISBN=978-3-030-83356-5. (2021).
[17] Lundberg, S. M. and Lee, S. A Unified Approach to Interpreting Model Predictions. 31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA. (2017).
[18] Lundberg, S. M., Erion, G., Chen, H., DeGrave, A., Prutkin, J. M., Nair, B., Katz, R., Himmelfarb, J., Bansal, N. and Lee, S. Explainable AI for Trees: From Local Explanations to Global Understanding. arXiv:1905.04610 (2019).
[19] Mothilal, R. K., Sharma, A. and Tan, C. Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations. CoRR, Volume: abs/1905.07697. URL: https://arxiv.org/abs/1905.07697. (2019).
[20] Markus, A. F., Kors, J. A. and Rijnbeek, P. R. The role of explainability in creating trustworthy artificial intelligence for health care: A comprehensive survey of the terminology, design choices, and evaluation strategies. Journal of Biomedical Informatics, Volume 113, 103655, ISSN 1532-0464, https://doi.org/10.1016/j.jbi.2020.103655. (2021).
[21] Maloney, K. O., Buchanan, C., Jepsen, R. D., Krause, K. P., Cashman, M. J., Gressler, B. P., Young, J. A., Schmid, M. Explainable machine learning improves interpretability in the predictive modeling of biological stream conditions in the Chesapeake Bay Watershed, USA. Journal of Environmental Management, Volume 322, 2022, 116068, ISSN 0301-4797, https://doi.org/10.1016/j.jenvman.2022.116068. (2022).
[22] Meo, R., Nai, R. and Sulis, E. Explainable, Interpretable, Trustworthy, Responsible, Ethical, Fair, Verifiable AI... What’s Next?. In: Chiusano, S., Cerquitelli, T., Wrembel, R. (eds) Advances in Databases and Information Systems. ADBIS 2022. Lecture Notes in Computer Science, vol 13389. Springer, Cham. https://doi.org/10.1007/978-3-031-15740-0_3 (2022).
[23] Nandi, A. and Pal, A. K. Interpreting Machine Learning Models: Learn Model Interpretability and Explainability Methods. Apress Berkeley, CA. ISBN= 978-1-4842-7801-7, DOI= https://doi.org/10.1007/978-1-4842-7802-4. (2021).
[24] Pearl, J. Causality. Cambridge university press. ISBN= 9780511803161, DOI= https://doi.org/10.1017/CBO9780511803161. (2009).
[25] Peters, J., Janzing, D. and Scholkopf, B. Elements of Causal Inference: Foundations and Learning Algorithms. The MIT Press. ISBN: 978-0262037310. (2017).
[26] Poon, A. I. F., and Sung, J. J. Y. Opening the black box of AI-Medicine. Journal of Gastroenterology and Hepatology, 36: 581– 584. https://doi.org/10.1111/jgh.15384. (2021).
[27] Ribeiro, M. T., Singh, S. and Guestrin, C. "Why Should I Trust You?": Explaining the Predictions of Any Classifier. arXiv: 1602.04938 (2016).
[28] Roscher, R., Bohn, B., Duarte, M. F. and Garcke, J. Explainable Machine Learning for Scientific Insights and Discoveries. in IEEE Access, vol. 8, pp. 42200-42216, 2020, doi: 10.1109/ACCESS.2020.2976199. (2020).
[30] Shapley, L. S. A Value for N-Person Games. Contributions to the Theory of Games, 307–317. (1953).
[31] Saeed, W. and Omlin, C. Explainable AI (XAI): A systematic meta-survey of current challenges and future opportunities. Knowledge-Based Systems, Volume 263, 110273, ISSN 0950-7051, https://doi.org/10.1016/j.knosys.2023.110273. (2023).
[32] Verma, P. Data Efficient Algorithms and Interpretability Requirements for Personalized Assessment of Taskable AI Systems. International Joint Conference on Artificial Intelligence. (2021).
[33] Wachter, S. and Mittelstadt, B. and Russell, C. Counterfactual Explanations without Opening the Black Box: Automated Decisions and the GDPR. arXiv:1711.00399. (2018).
[34] White, A. and Garcez, A. Measurable Counterfactual Local Explanations for Any Classifier. arXiv. DOI= 10.48550/ARXIV.1908.03020. Available from https://arxiv.org/abs/1908.03020. (2019).
[35] Xu, P., Ji, X., Li, M. and Lu, W. Small data machine learning in materials science. npj Comput Mater 9, 42. https://doi.org/10.1038/s41524-023-01000-z. (2023).