機器學習是大數據時代下資料科學快速發展的研究課題，其應用範圍含括在工程、生醫、社會經濟與工程領域，在加速解決許多既有問題時，也創造出更多的應用。數據科學包含數學與統計方法，計算機軟硬體，以及應用領域知識等三個面向，掌握數學科學分析工具是大數據研究的基本技能，而了解其背後數學與統計的基本原理，可更有效率針對資料特性去選擇正確的模型進行分析。
本課程以Python套件Scikit-learn作為機器學習軟體實作入門，其內容含括：
1.監督式學習
(1)迴歸模型：簡單線性迴歸，多變數線性迴歸，多項式迴歸
(2)分類模型：邏輯迴歸，支持向量機，貝氏分類器，決策樹，隨機森林
2.非監督式學習
(1)分群模型：K-means (2) 特徵降微：主成分分析
3.機器學習實作上會面臨的問題與解決的辦法，包含資料前處理、超參數調教、偏差與變異、欠擬合、過度擬合以及學習曲線、測試曲線的行為。
此外，在有了機器學習基本概念後，我們將學習由Keras實作人工神經網路（ANN）、深度學習（DNN）與卷積神經網路（CNN），包含迴歸分析、手寫數字辨識、影像識別等人工智慧實務應用，並了解其背後直觀的數學意義與技巧。Machine learning is a rapidly developing research topic in data science in the era of big data. Its application scope includes engineering, biomedicine, social economy and engineering fields. It accelerates the solution of many existing problems and creates more applications. Data science includes three aspects: mathematical and statistical methods, computer software and hardware, and application domain knowledge. Mastering mathematical scientific analysis tools is a basic skill for big data research, and understanding the basic principles of mathematics and statistics behind it can be more efficient. Select the correct model for analysis based on the characteristics of the data.
This course uses the Python package Scikit-learn as an introduction to machine learning software implementation. Its content includes:
1. Supervised learning
(1) Regression model: simple linear regression, multivariable linear regression, polynomial regression
(2) Classification models: logistic regression, support vector machine, Bayesian classifier, decision tree, random forest
2. Unsupervised learning
(1) Grouping model: K-means (2) Feature reduction: Principal component analysis
3. Problems and solutions faced in machine learning implementation, including data pre-processing, hyperparameter tuning, deviation and mutation, under-fitting, over-fitting, and the behavior of learning curves and test curves.
In addition, after understanding the basic concepts of machine learning, we will learn how to implement artificial neural networks (ANN), deep learning (DNN) and convolutional neural networks (CNN) using Keras, including regression analysis, handwritten digit recognition, and image processing. Identify and other practical applications of artificial intelligence, and understand the intuitive mathematical meaning and techniques behind them.
Machine learning is the science of data analysis that enables computers to learn without being explicitly programmed. From the computer science point of view, unlike computational statistics dealing with prediction-making or data mining focusing on data-exploring, machine learning uses data to iteratively detect patterns and adjust models accordingly. This introductory course provides students an overview of the field of machine learning, as well as of its fundamental concepts and algorithms from practical perspective. Usually, machine learning algorithms are categorized as being supervised or unsupervised. Some of the important topics include (1) Supervised learning (Linear and Logistic Regressions, Classification and Regression Trees, Support Vector Machines, and Neural Networks). (2) Unsupervised learning (Association Rules and Cluster Analysis). (3) Others (Boosting and Random Forests).
Machine learning is the science of data analysis that enables computers to learn without being explicitly programmed. From the computer science point of view, unlike computational statistics dealing with prediction-making or data mining focusing on data-exploring, machine learning uses data to iteratively detect patterns and adjust models accordingly. This introductory course provides students an overview of the field of machine learning, as well as of its fundamental concepts and algorithms from practical perspective. Usually, machine learning algorithms are categorized as being supervised or unsupervised. Some of the important topics include (1) Supervised learning (Linear and Logistic Regressions, Classification and Regression Trees, Support Vector Machines, and Neural Networks). (2) Unsupervised learning (Association Rules and Cluster Analysis). (3) Others (Boosting and Random Forests) .
Ian Goodfellow, Yoshua Bengio and Aaron Courville, Deep Learning, 2016,The MIT Press
Ian Goodfellow, yobrushB鞥IO and Aaron Cville, deep learning, 2016, the MIT press
| 評分項目 Grading Method | 配分比例 Grading percentage | 說明 Description |
|---|---|---|
| 隨堂測驗隨堂測驗 Quiz in class |
15 | |
| 作業作業 Homework |
55 | |
| 期末報告期末報告 Final report |
20 |
