数学书, 摘录自Lin Dahua的博客
Algebra
线性代数:subspace,orthorgonality, Eigenvalues and eigenvectors, linear transform
Topology(拓扑学)
帮助理解基础概念 open set,closed set,set basis,Hausdauf,continuous function,metric space,Cauchy sequence,neighborhood,compactness,connectivity等.
Topology (2dn Ed.), by James Munkres.
Analysis, and Geometry
微积分:分析的基础
偏微分方程(Partial Differential Equation):描述动态过程; 在vision中用来描述连续场的运动或扩散过程,如Level set,Optical flow
实分析:一切分析的基础,
复分析
泛函分析(Functional Analysis):微积分从有限维空间到无限维空间的拓展;希尔伯特空间理论和算子谱论在机器学习领域有核心作用。Kernel Learning,Gaussian Process,Kernel(Inner product)。Introductory Functional Analysis with Applications, by Erwin Kreyszig.
调和分析:用于信号处理和分解,比如傅里叶变换,compressed sensing, sparse coding
微分方程理论:连续时间的分布变化,
李群,李代数理论:动态过程建模,随即过程。映射,变换,度量,划分等在研究learning中的代数方法有作用。
Lie Groups, Lie Algebras, and Representations: An Elementary Introduction. by Brian C.Hall.
图论(Graph theory)
Optimization and Numerical Computation
数值优化:Gradient descent
凸优化(Convex Optimization)
组合优化(Combinatorial Optimization)
线性优化(Linear Programming):基于图和网络的优化问题
Probability Theory
现代概率理论:基础是测度理论(Measure theory);积分理论;independence,conditional expectation,martingale,可测性
Statistics
Applied Multivariate Statistical Analysis (5th Ed.) , by Richard A. Johnson and Dean W.Wichern;
Introduction to Graphical Models (draft version) , by M. Jordan and C. Bishop. :贝叶斯模型,统计网络的估计和推断
Stochastic Processes
随机过程:随机函数,Markov,Brownian motion,Gaussian Process,Martingale,Poisson Process