GraphPKU / MachineLearning2024

67 stars 2 forks source link

The syllabus of the course #8

Closed Cgfyufsygsm closed 2 weeks ago

Cgfyufsygsm commented 2 weeks ago

Dear teacher and TAs, I wonder if there is a syllabus for this course? I couldn't find one in course.pku.edu.cn. Thank you!

muhanzhang commented 2 weeks ago

Here is a tentative syllabus.

<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40">

1 线性回归(Linear Regression) 3 经验风险最小化, 矩阵求导,线性回归闭式解,岭回归和Lasso,最小二乘几何解释 ERM, Matrix Derivatives, Closed-form solution to Linear Regression, Ridge Regression and Lasso, Geometric view of Least Squares
2 逻辑回归(Logistic Regression) 6 最大似然原则,交叉熵损失,线性可分,逻辑回归凸函数的证明,平方损失函数分类的缺点,Softmax Regression,线性回归的最大似然解释,岭回归的最大后验解释 MLE, Cross Entropy, Linear Separable, Proof of the convexity of Logistic Regression, Drawbacks of squared loss for classification, Softmax Regression, MLE of Linear Regression, MAP of Linear Regression
3 偏差-方差分解(Bias-Variance Decomposition) 3 偏差方差分解,模型选择,奥卡姆剃刀原理 Bias-Variance Decomposition, Model Selection, Occam’s Razer
4 支持向量机与对偶理论(Support Vector Machine and Dual Theory) 9 约束优化问题,拉格朗日乘子法,KKT条件,支持向量机主形式,对偶形式,对偶问题的SMO算法,核方法,松弛变量 Constrained optimization, Lagrange Method, KKT conditions, Primal and Dual forms of SVM, SMO, Kernel methods, Slack variables
5 表示定理(Representer’s Theorem) 3 再生核希尔伯特空间,表示定理及证明,利用表示定理解释岭回归 RKHS, Representer’s Theorem, Revisiting Ridge Regression
6 学习理论(Learning Theory) 3 PAC理论,成长函数,VC维,VC泛化界 PAC Learning Theory, Growth function, VC-dimension, VC Generalization Bound
7 树模型和集成学习(Tree Models and Ensemble Learning) 6 信息熵,信息增益,增益率,基尼系数,决策树,回归树,连续特征,Bagging, 随机森林,Boosting, AdaBoost, GBDT Entropy, Information gain, Information gain ratio, Gini-index, Decision Tree, Bagging, Random Forest, Boosting, AdaBoost, GBDT
8 高斯过程(Gaussian Process) 3 多元高斯分布,随机过程,高斯过程回归,贝叶斯优化 Multivariate Gaussian Distribution, Random Process, Gaussian Process Regression, Bayesian Optimization
9 图模型基础(Graphical Models Basics) 3 贝叶斯网络,条件独立性,D-separation,无向图模型 Bayesian networks, Conditional independence, D-separation, Unconditional graphical models
10 无监督学习(Unsupervised Learning) 9 主成分分析,混合高斯模型,EM算法,变分自编码器,扩散模型 PCA, Mixture of Gaussians, Expectation-Maximization, VAE, Diffusion Models

Cgfyufsygsm commented 2 weeks ago

Thank you for your providing!