Update syllabus：2005/7/12

Update lecture notes : 2005/7/12

Access Index :

Spring Semester

- Outline of lecture
- The objective of this course is to introduce basic ideas and practical methods of discovering useful structure hidden in the data.
- Purpose of lecture
- The objective of this course is to introduce basic ideas and practical methods of discovering useful structure hidden in the data.
- Plan of lecture
- 1. Introduction

2. Pseudo Biorthogonal Basis

3. Principal Component Analysis

4. Kernel Principal Component Analysis

5. Non-Gaussian Component Analysis

6. Spectral Methods of Dimensionality Reduction

7. K-means Clustering

8. Spectral Clustering

9. Outlier Detection

10. Kernel Outlier Detection

11. Independent Component Analysis

12. Blind Source Separation

13. Concluding Remarks and Future Prospects - Textbook and reference
- None. Handouts are distributed if necessary.
- Related and/or prerequisite courses
- Probability and Statistics, Pattern Recognition
- Evaluation
- Reports related to data analysis and students' projects.
- Comments from lecturer
- In order to really learn the methods, it is important to actually use them. Analyzing your own data using the learned methods is expected.
- Supplement
- [Office Hours]

Anytime if available.