Gilbert Strang
Matrix Methods in Data Analysis — thirty-six lectures from MIT, Spring 2018.Matrix Methods in Data Analysis, Signal Processing, and Machine Learning. Thirty-six lectures from MIT’s Spring 2018 course.
Matrix Methods in Data Analysis · 36 lectures
Course Introduction of by Professor Strang
An Interview with Gilbert Strang on Teaching Matrix Methods
1: The Column Space of A Contains All Vectors Ax
2: Multiplying and Factoring Matrices
3: Orthonormal Columns in Q Give Q’Q = I
4: Eigenvalues and Eigenvectors
5: Positive Definite and Semidefinite Matrices
6: Singular Value Decomposition (SVD)
7: Eckart-Young: The Closest Rank k Matrix to A
8: Norms of Vectors and Matrices
9: Four Ways to Solve Least Squares Problems
10: Survey of Difficulties with Ax = b
11: Minimizing x Subject to Ax = b
12: Computing Eigenvalues and Singular Values
13: Randomized Matrix Multiplication
14: Low Rank Changes in A and Its Inverse
15: Matrices A(t) Depending on t, Derivative = dA/dt
16: Derivatives of Inverse and Singular Values
17: Rapidly Decreasing Singular Values
18: Counting Parameters in SVD, LU, QR, Saddle Points
19: Saddle Points Continued, Maxmin Principle
20: Definitions and Inequalities
21: Minimizing a Function Step by Step
22: Gradient Descent: Downhill to a Minimum
23: Accelerating Gradient Descent (Use Momentum)
24: Linear Programming and Two-Person Games
25: Stochastic Gradient Descent
26: Structure of Neural Nets for Deep Learning
27: Backpropagation: Find Partial Derivatives
30: Completing a Rank-One Matrix, Circulants!
31: Eigenvectors of Circulant Matrices: Fourier Matrix
32: ImageNet is a Convolutional Neural Network (CNN)
33: Neural Nets and the Learning Function
34: Distance Matrices, Procrustes Problem
35: Finding Clusters in Graphs
36: Alan Edelman and Julia Language