Traditionals

some traditional machine learning algorithms

Survey Papers / Repos

• Top 10 algorithms in data mining. [ICDM'06]

• josephmisiti/awesome-machine-learning

Resources

• Coursera Machine Learning by Andrew Ng

Tasks

Supervised

• Linear Regression

$$y=ax+b\\ L(y,\hat{y}) = (y-\hat{y})^2$$

• Logistic Regression

$$y=\frac{1}{1+e^{-(ax+b)}} \\ L(y,\hat{y}) = -\hat{y}\log y - (1 - \hat{y}) \log (1-y)$$

• Naive Bayes

$$P(A|B) = \frac{P(B|A)P(A)}{P(B)}$$

• Support Vector Machine (SVM)

  • Training process: Lagrange -> Dual Problem -> SMO

  $$\min \frac{1}{2} ||w||^2 \\ \text{s.t.}~y^{(i)}(w^{T}x^{(i)}+b) \geq 1, i=1,...,m$$
• K Nearest Neighbor (kNN)

• Expectation-Maximization (EM)

• Linear Discrimant Analysis (LDA)

• Decision Tree

• Random Forest

• Gradient Boosting Tree (GBDT)

Semi-supervised

Weakly-supervised

Unsupervised

• Clustering

  • K-means

  • Mean-shift

  • DBSCAN

• Principal Component Analysis (PCA)

• Latent Dirichlet allocation (LDA) Topic Modeling

Others

Ensemble

• K-Fold Cross Validation

• Bagging

• Boosting

Metrics

	True Samples	False Samples
Predict True	True Positive	False Positive [Type I Error]
Predict False	False Negative [Type II Error]	True Negative

• Precision and Recall

  • $\text{Precision} = \frac{\text{TP}}{\text{TP} +\text{FP}}$
  • $\text{Recall} = \frac{\text{TP}}{\text{TP}+\text{FN}}$
• F1 Score

  • $\text{F1 score} = 2 \cdot\frac{\text{Precision} \cdot \text{Recall}}{\text{Precision} +\text{Recall}}$
• Receiver Operating Characteristic (ROC)

  • $\text{TPR} = \frac{\text{TP}}{\text{TP}+\text{FN}}$
  • $\text{FPR} = \frac{\text{FP}}{\text{FP}+\text{TN}}$
• Area Under ROC (AUC)

• Confusion Matrix

Reference

• SVM: https://www.cnblogs.com/jerrylead/archive/2011/03/13/1982639.html