課程名稱 |
離散幾何分析 Discrete Geometry Analysis |
開課學期 |
108-2 |
授課對象 |
理學院 數學研究所 |
授課教師 |
崔茂培 |
課號 |
MATH5241 |
課程識別碼 |
221 U8780 |
班次 |
|
學分 |
2.0 |
全/半年 |
半年 |
必/選修 |
選修 |
上課時間 |
星期三8,9(15:30~17:20) |
上課地點 |
天數101 |
備註 |
總人數上限:30人 |
Ceiba 課程網頁 |
http://ceiba.ntu.edu.tw/1082MATH5241_ |
課程簡介影片 |
|
核心能力關聯 |
本課程尚未建立核心能力關連 |
課程大綱
|
為確保您我的權利,請尊重智慧財產權及不得非法影印
|
課程概述 |
Manifold learning encompasses much of the disciplines of geometry, computation, and statistics, and has become an important research topic in data mining and statistical learning. The simplest description of manifold learning is that it is a class of algorithms for recovering a low-dimensional manifold embedded in a high-dimensional ambient space. This course aims to help those who want to understand the geometric aspects of various learning algorithms.
This course focuses on introducing spectral methods for the dimensional reduction problem. Topics include (1) Principal component analysis; (2) Multidimensional scaling; (3) Locally linear embedding; (4) Maximum variance unfolding; (5) Diffusion map and (6) Cluster analysis.
We will introduce the theories and implementations of these topics.
|
課程目標 |
Students are able to understand the theories and implementations of
modern dimension reduction methods in manifold learning and apply
manifold learning theories to data analysis. |
課程要求 |
The prerequisite of this mini-course is linear algebra and calculus. |
預期每週課後學習時數 |
|
Office Hours |
|
指定閱讀 |
Ten Lectures on Diffusion Map by Amit Singer and Hau-tieng Wu |
參考書目 |
Ten Lectures and Forty-Two Open Problems in the Mathematics of Data Science
available at https://ocw.mit.edu/courses/mathematics/18-s096-topics-in-
mathematics-of-data-science-fall-2015/lecture-notes/MIT18_S096F15_TenLec.pdf
Geometric Structure of High-Dimensional Data and Dimensionality Reduction by
Jianzhong Wang |
評量方式 (僅供參考) |
|
週次 |
日期 |
單元主題 |
第5週 |
4/01 |
1. Finish the topics of PDS Kernel
(THEOREM 1.29. p17 I made a mistake in one step.
i.e. L(x,y)=1 for for all x, y is a PDS kernel
L_{i,j}=1 not L_{ij}=\delta_{ij}.
2. Construct a useful PDS kernel (PROPOSITION 1.30) and a transition matrix Lemma 1.31)
3. Show that diffusion distance is a metric if there is no zero eigenvalue. PROPOSITION 1.21 p 13
4. Discuss the ideas of PCA (p19,-p20) |
第6週 |
4/08 |
Discuss PCA and MDS. |
第15週 |
6/10 |
The final project presentation will be on June 24th.
You should turn in the final project by July 1st. |
|