学习目标
线性代数作为现代数学的核心，是理解、设计和实现机器学习算法的基础，特别是在深度学习及神经网络的应用中显得尤为重要。本次学习小组将通过共同学习Gilbert Strang教授的公开课MIT18.065 Matrix Methods in Data Analysis, Signal Processing, and Machine Learning, 以及对应的教材Linear Algebra and Learning from Data，对Machine Learning里涉及的主要线性代数相关知识，如数值现代，最优化，概率和统计进行全面的梳理，力图从现代的角度来理解流行的算法。
学员组成:
助教：一名（已到位）
领队：一名（已到位）
组员：8人（含领队），不限专业，但要求有责任心，爱表达提问，有独立思考和钻研的能力。
报名要求:
课程时间:
琪石学分: 0
部分学习大纲:
Week 1: Basics of Linear Algebra (Lectures 1-4), Column Space，Matrix Factorization，Orthogonality，Eigenvalues and Eigenvectors.
Week 2: Highlights of Linear Algebra (Lectures 5-8), Positive Definite Matrices and Semidefinite Matrices，SVD & PCA，the Best Low，Rank Matrix，Norms of Vectors and Functions and Matrices.
Week 3: Applications and Computations (Lectures 9-13), Least Squares Problems，Numerical Linear Algebra， Computing Eigenvalues and Eigenvectors，Randomized Matrices.
Week 4: Matrix Analysis (Lectures 14-17), Low Rank Changes in A and Its Inverse，Derivatives of Matrices and Inverses，More on Singular Values.
Week 5: Matrix Optimization I (Lectures 18-22), Saddle Points and Maximum Principle，Minimization and Gradient Descent method.
Week 6: Matrix Optimization II (Lectures 23-25), Accelerating Gradient Descent， Linear Programming，Stochastic Gradient Descent.
Week 7: Learning from Data I (Lectures 26-27, 30-31), Neural Networks & Backpropagation，Circulants and Fourier Matrices.
Week 8: Learning from Data II (Lectures 32-35), Convolutional Neural Nets，Procrustes Problem，Clustering.
学习资料:
公开课：https://ocw.mit.edu/courses/18-065-matrix-methods-in-data-analysis-signal-processing-and-machine-learning-spring-2018/
教材：https://math.mit.edu/~gs/learningfromdata/
...
Activity type: OnlineStudyGroups
...
Announce date: Jan. 9, 2023
Registration deadline: May 7, 2023
Start date: April 30, 2023
End date: June 20, 2023
...
Leader: Not disclose
Maximum participants: 50
Maximum Applicants: 50
7 people already applied
...
Fee for premium members: 50.0
Fee for all others: 100.0
Copyright 2020 by Flash Kick In Consulting Inc. All Rights Reserved.