機器學習

Machine Learning

學期

112-1

學分

3 學分

當期課號

639004

永久課號

AICA30009

開課單位

智慧科學暨綠能學院

授課教師

馬清文

校區

歸仁

類別

選修

上課時間表

節	週二
5 13:20–14:10	機器學習 CM212 3 節連堂
6 14:20–15:10
7 15:30–16:20

* 根據陽明交大上課時間表所列

概述

This course, with the assistance of a well-known free online video, Stanford cs229 machine learning summer edition, provides basic, general, and advanced machine learning concepts. Topics include supervised learning, reinforcement learning, unsupervised learning, variational inference, etc. We will also include decision trees and recent advanced topics, such as self-supervised learning, contrastive learning, and large-language models. Classes will be conducted either remotely or in-person.. For remote classes, we use Microsoft Teams with the link https://tinyurl.com/5trweptk

先修科目

Basics of probability theory, linear algebra, and multivariable calculus Reasonably computer programming skills in Matlab/Python/numpy.

教學方式

Reference web sites: https://cs229.stanford.edu/syllabus-summer2020.html https://github.com/maxim5/cs229-2018-autumn https://www.youtube.com/playlist?list=PLblh5JKOoLUICTaGLRoHQDuF_7q2GfuJF

評分方式

5 Homework Problem sets: 1 Term project proposal: 1 Term project report = 40%:20%:40%

週次計畫

週次	主題
第 1 週	Part 1. Basic Machine learning concepts 1. Course outline, Introduction and Logistics K-near neighbors, Decision tree, Random forest https://tinyurl.com/5trweptk
第 2 週	2. Gradient Boost Decision tree, XG-Boost, Tabular data vs. multi-media data CM212 or https://tinyurl.com/5trweptk
第 3 週	* Supervised Learning * 3. Linear Regression [Stochastic] Gradient Descent ([S]GD) Normal Equations Probabilistic Interpretation Maximum Likelihood Estimation (MLE) Problem Set 1 and 2 Release CM212 or https://tinyurl.com/5trweptk
第 4 週	4. Perceptron Logistic Regression Newton's Method CM212 or https://tinyurl.com/5trweptk
第 5 週	Part 2. General Machine Learning Concepts 5. Exponential Family Generalized Linear Models (GLM) Gaussian Discriminant Analysis (GDA) Naive Bayes Laplace Smoothing CM212 or https://tinyurl.com/5trweptk
第 6 週	6. Kernel Methods Support Vector Machine CM212 or https://tinyurl.com/5trweptk
第 7 週	7. Support Vector Machine Application Bayesian Methods (optional) Parametric (Bayesian Linear Regression, optional) Non-parametric (Gaussian process, optional) CM212 or https://tinyurl.com/5trweptk
第 8 週	8. Neural Networks and Deep Learning Problem set 3 Release CM212 or https://tinyurl.com/5trweptk
第 9 週	* Theory * 9. Bias and Variance Regularization, Bayesian Interpretation Model Selection CM212 or https://tinyurl.com/5trweptk
第 10 週	10. Bias-Variance tradeoff (wrap-up) Empirical Risk Minimization Uniform Convergence CM212 or https://tinyurl.com/5trweptk
第 11 週	* Reinforcement Learning * 11. Reinforcement Learning (RL) Markov Decision Processes (MDP) Value and Policy Iterations Learning MDP model Continuous States Problem set 4.2 4.4 release CM212 or https://tinyurl.com/5trweptk
第 12 週	12. Recap, Review, and Term project proposal CM212 or https://tinyurl.com/5trweptk
第 13 週	* unsupervised learning * 13. K-means clustering Mixture of Gaussians (GMM) Expectation Maximization (EM) Principal Components Analysis (PCA) Independent Components Analysis (ICA) Problem set 4.1 4.3 release CM212 or https://tinyurl.com/5trweptk
第 14 週	Part3. Advanced machine learning concepts 14. Variational Inference EM Variants Variational Autoencoder Principal Components Analysis (PCA) CM212 or https://tinyurl.com/5trweptk
第 15 週	15. Advanced topics To be determined. CM212 or https://tinyurl.com/5trweptk
第 16 週	16. Term project report CM212 or thttps://tinyurl.com/5trweptk

教科書

https://cs229.stanford.edu/lectures-spring2022/main_notes.pdf

Office Hours

地點: online
時間: on appointment
聯絡方式: email: machingwen@ncyu.edu.tw