serendipity-55428/nctu-2020ml

2020 NCTU Machine Learning (交大機器學習-簡仁宗)

2020 NCTU Machine Learning

Text Book: PATTERN RECOGNITION and MACHINE LEARNING (BISHOP)

HomeWork

HW 1

• Bayesian Linear Regression (手寫)
• Linear Regression

HW 2

• Sequence Bayesian Regression
• Logistic Regression

HW 3

• Gaussian Process
• Support Vector Machine
• Gaussian Minxtures Model

Midterm and Final Exam

考試規則: 可以帶一張 A4 大抄 (雙面皆可使用)

準備重點: 字抄小一點 (PS.帶放大鏡?)

Midterm

1. Information Theory

   • 證明 Convex function 的性質 (tips: Jensen's Inequality) (text book P.57)
   • 說明為什麼 KL-divergence 恆大於 0
   • Kullback-Leibler divergence between the joint distribution (text book P.57)

2. 習題 1.31 (text book P.65) (PS.這題不確定)

3. 習題 2.5 (text book P.128)

4. Exponential family and sufficient statistics

   • 用指數族特性找出 Bernoulli 和 Gaussian 的充份統計量
   • Maximum likelihood and sufficient statistics (text book P.116)

5. The Evidence Approximation

   • Evaluation of the evidence function (text book P.166、167)
   • Maximizing the evidence function (text book P.168、169)

6. Probabilistic Generative Models Maximum likelihood solution (text book P.200、201)

Final

因為期末考卷沒有發回來,下面列出我有印象的考題:

1. Nadaraya-Waston Model

2. Support Vector Machine

   • Soft-Margin 、 Hard-Margin 兩者有何差異

3. EM Algorithm

   • EM Algorithm in General(text book 9.4)

======= 以上配分 75 % =======

4. Logistic Regression (手寫code) 25%