This repository implements some Classical machine learning algorithms in Python (with NumPy and SciPy).
Comments and discussion are always highly appreciated :)
- k-NN
- Perceptron
- Classification Decision Tree (ID3/C4.5)
- Classification Tree (CART Algorithm)
- Regression Tree (CART Algorithm)
- Support Vector Machine (SMO Algorithm)
- Feedforward Neural Network (MLP)
- KMeans Clustering
- Gaussian Mixture Model Clustering (by EM algorithms/Gibbs Sampling/Variational Inference)(More methods can be found in this repository)
- Hierarchical Clustering (implementing multiple distances: Euclidean/Minkowski/Manhattan/Chebyshev/Manhalanobis/Correlation/Cosine)
- Principle Component Analysis
- Kernel PCA
- Factor Analysis(Linear Gaussian Model with anisotropic variances)
- Probabilistic PCA(Number of PCs are automatically determined by ARD prior)
- PPCA by Pyro(Illustrated in Jupyter Notebook)
该仓库实现了一些经典的机器学习算法, 欢迎讨论和批评指正!
- k近邻
- 感知机
- 分类决策树(ID3/C4.5)
- 分类树(CART算法)
- 回归树(CART算法)
- 支持向量机(SMO算法)
- 前馈神经网络(MLP)
- K均值聚类
- 高斯混合模型聚类(利用EM算法/Gibbs采样/变分推断)(更多方法见此仓库)
- 层次聚类(实现多种距离:欧氏距离/闵可夫斯基距离/曼哈顿距离/切比雪夫距离/马哈拉诺比斯距离/相关系数相似度/余弦相似度)
- 主成分分析
- 核主成分分析
- 因子分析(各向异性方差线性高斯模型)
- 概率主成分分析(利用ARD先验自动确定主成分数量)
- Pyro实现的概率主成分分析(Jupyter Notebook)