This course aims at equipping students with the theory foundation and the best

practice (as well as applications) of machine learning. The course contains four

major parts:

1. Theory foundation and algorithms of machine learning

2. Theory and practice of deep learning

3. Semester-long projects

4. Paper presentation

This course is designed to provide a graduate-level student a thorough

understanding in the methodologies, technologies, mathematics and algorithms

currently required by people who do research in machine learning. After attending

this course, students are expected to (a) have a solid grasp on machine learning

theories, (b) earn hands-on experience in using machine learning for real-world

problems, (c) be capable of catching up the latest machine learning literatures

and trends.

The lectures will be provided in Mandarin, and most of the teaching and research

materials will be in English. The materials and directions of this course are

tentative and subject to change.

1. Introduction

2. Statistical foundation

3. Convex optimization

4. Learning theory

5. Algorithms: decision tree

6. Algorithms: logistic regression, naive bayes classifier

7. Algorithms: support vector machine

8. Deep learning concept

9. Feedforward neural network (FFNN)

10. Basic components: activation function, loss function, etc

11. Techniques to fight overfitting: dropout and regularization

12. Minibatch and AdaGrad

13. Convolutional neural network (CNN)

14. Convolution filter, and max-pooling

15. CNN variants: res-net and others

16. Recurrent neural network (RNN) and Long-Short-Term Memory (LSTM)

17. em

18. Deep Learning applications

Grading

R26; 5%: Paper Presentation

R26; 25%: Taking lecture notes with Latex

R26; 25%: Homework or pop quizzes

R26; 45%: Semester-long projects

R26; 無故不到2次成績不及格