機器學習技術與應用

課程大綱

The course begins with the machine learning workflow, including data, features, labels, training, validation, testing, and model evaluation. Students then study major classification methods, including kNN, Naïve Bayes, discriminant analysis, decision trees, SVMs, ensemble methods, and neural networks. Regression topics include linear, robust and regularized regression, support vector regression, regression trees, ensembles, Gaussian processes, and neural networks. The course further covers clustering methods, dimensionality reduction, feature selection, and feature extraction from raw and time-varying signals. Deep learning is introduced through neural representations, transfer learning, training curves, and model limitations. The semester concludes with an evidence-based team symposium and an industry expert workshop on reinforcement learning, while class activities emphasize controlled comparison, evidence interpretation, peer questioning, and English technical communication.