2 項進行中

115-1 選課時程

進行中

  • 初選第一階段 6/15 – 6/18
  • 初選第二階段 6/22 – 6/25
  • 校際選修 進行中 8/24 – 9/18
  • 初選第三階段 8/31 – 9/3
  • 開學後加退選 進行中 9/7 – 9/21
  • 逾期加退選 9/21 – 9/24
選課資源

加入行事曆

選擇訂閱 Google Calendar,或下載通用的 ICS 檔案。

使用 Google Calendar 時,Google 會收到這份課表的公開連結。

數據科學矩陣方法(英文授課)

Matrix Methods in Data Science

學期
108-1
學分
0 學分
當期課號
5948
永久課號
IDS5009
開課單位
數據科學與工程研究所碩士班
授課教師
彭文孝
校區
光復
類別
選修
上課時間表
週二
週四
3
10:10–11:00
數據科學矩陣方法(英文授課)
EC122(光復)
2 節連堂
4
11:10–12:00
8
16:30–17:20
數據科學矩陣方法(英文授課)
EC122(光復)

* 根據陽明交大上課時間表所列

概述

This course extends matrix methods in Linear Algebra to cover their applications to data science, including data analysis, signal processing, and machine learning. It shall equip students with the required matrix methods on which data science depends.

先修科目

1. Linear Algebra

備註

無備註

教學方式

教師未提供此項資料

評分方式

Mid-term x 2 -- 50% Final x 1 -- 30 % Homework (including computer assignments) – 20% Project (??)

課程大綱
  • Highlights of Linear Algebra

    1. The four fundamental subspaces 2. Orthogonal matrices and subspaces 3. Eigenvalues and eigenvectors 4. Symmetric positive definite matrices 5. Singular value and singular vectors in SVD 6. Principle components and best low rank matrices 7. Rayleigh quotients and generalized eigenvalues 8. Norms of vectors, functions, and matrices 9. Factoring matrices and tensors

  • Computations with Large Matrices

    1. Least squares 2. Three bases for the column space 3. Randomized linear algebra

  • Low Rank and Compressed Sensing

    1. Changes in the inverse of matrices from changes in them 2. Interlacing eigenvalues and low rank signals 3. Rapidly decaying singular values 4. Compressed sensing and matrix completion

  • Special Matrices

    1. Fourier transforms 2. Shift matrices and circulant matrices 3. The kronecker product 4. Sine and Cosine transforms from Kronecker sums 5. Toeplitz matrices and shift invariant filters 6. Graphs and Laplacians 7. Clustering by spectral methods and k-means 8. Completing rank one matrices 9. The orthogonal procrustes problem 10. Distance matrices

  • Optimization

    1. Minimum problems: convexity and Newton’s Method 2. Lagrange multipliers 3. Linear programming, game theory, and duality 4. Gradient descent 5. Stochastic gradient descent and ADAM

  • Learning form data

    1. The construction of deep neural networks 2. Convolutional neural nets 3. Backpropagation and the chain rule 4. Hyperparameters

週次計畫
週次主題
第 1 週

Highlights of Linear Algebra

第 2 週

Highlights of Linear Algebra

第 3 週

Highlights of Linear Algebra

第 4 週

Computations with Large Matrices

第 5 週

Computations with Large Matrices

第 6 週

Computations with Large Matrices

第 7 週

Low Rank and Compressed Sensing

第 8 週

Low Rank and Compressed Sensing

第 9 週

Low Rank and Compressed Sensing

第 10 週

Special Matrices

第 11 週

Special Matrices

第 12 週

Special Matrices

第 13 週

Optimization

第 14 週

Optimization

第 15 週

Optimization

第 16 週

Learning form data

第 17 週

Learning form data

第 18 週

Learning form data

教科書

Text: “Linear Algebra and Learning from Data,” Gilbert Strang, 1st Ed., published by Wellesley-Cambridge Press, 2019. Online resources: https://ocw.mit.edu/courses/mathematics/18-065-matrix-methods-in-data-analysis-signal-processing-and-machine-learning-spring-2018/

Office Hours
地點
教師未提供此項資料
時間
教師未提供此項資料
聯絡方式
教師未提供此項資料