Skip to content

Latest commit

 

History

History
178 lines (163 loc) · 5.72 KB

README.md

File metadata and controls

178 lines (163 loc) · 5.72 KB

tinyml

利用numpy实现的一些周志华《机器学习》(西瓜书)一书及 斯坦福cs229课程中的算法,宜配合西瓜书和cs229课件食用。并选择性实现了一些经典算法的简易版本, 如 按照陈天奇的slides实现的XGBRegressor。

已经实现的算法

和sklearn实现的比较

Algorithm vs. RMSE sklearn-boston
tinyml sklearn
LinearRegression 27.196 27.196
SGDRegressor 27.246 27.231
DecisionTreeRegressor 21.887 21.761
RandomForestRegressor 21.142 21.142
GradientBoostRegressor 16.778 16.106
XGBRegressor 20.149 15.7
Algorithm vs. RMSE sklearn-breast_cancer
tinyml sklearn
NaiveBayes 90.64% 90.64%
LogisticRegression 92.98% 92.98%
LDA 94.15% 92.40%
GDA 92.40% 93.57%
SVC 86.55% 92.98%
AdaboostClassifier 92.40% 92.40%
  • 聚类算法比较 代码
  • KMeans
tinyml KMeans sklearn KMeans
  • DBSCAN
tinyml DBSCAN sklearn DBSCAN
  • GMM
tinyml GMM sklearn GMM
  • AGNES
tinyml AGNES sklearn AGNES
  • 降维算法比较 代码
  • PCA
tinyml PCA sklearn PCA
  • KernalPCA
tinyml KernalPCA sklearn KernalPCA
  • LLE
tinyml LLE sklearn LLE
  • MDS
tinyml MDS sklearn MDS