Introduction to Machine Learning

Course Description

This course offers a comprehensive introduction to the foundational concepts and techniques of machine learning, equipping learners with the skills to analyze and solve real-world problems. Covering essential classical methods, the curriculum delves into classification, regression, clustering, and dimensionality reduction techniques, such as decision trees, support vector machines, k-means, and principal component analysis (PCA). With a balance of theory and practical implementation, participants will gain hands-on experience through coding exercises and case studies, ensuring they can apply machine learning concepts to a wide range of domains. Whether you’re a beginner or looking to solidify your understanding of classical approaches, this course provides a strong foundation for further exploration in the field of machine learning.

Syllabus

Foundations and Mathematics of Machine Learning:

  • Linear Algebra
  • Probability and Statistics
  • Maximum Likelihood and Maximum A posterior
  • Kernel Density Estimation
  • Optimization

Probabilistic Models:

  • Bayes Optimal Classifier
  • Naïve Bayes Classifier
  • Minimum Risk Classifier
  • Gaussian Discrimination Analysis (GDA)
  • KNN
  • Logistic Regression
  • Model Selection
  • Data Splitting
  • Performance Measures

Linear Models (Supervised Learning):

  • Perceptron and its variants
  • Least Square Methods
  • Fisher linear discriminant analysis
  • Support Vector Machine (SVM)
  • Linear Regression
  • Regularization (Ridge and LASSO)
  • Bias-Variance Trade-off

Non-Linear Models (Supervised Learning)

  • Basis Function Expansion
  • Radian Basis Function Network (RBF-NN)
  • Kernel Trick and SVM
  • Multi Layer Perceptron (MLP)
  • Decision Trees
  • Ensemble Learning
  • Bagging
  • Stacking
  • Boosting
  • Gradient Boosting Machine (GBM)
  • Gradient Tree Boosting (TreeBoost)
  • Extreme Gradient Boosting (XGBoost)

Unsupervised Learning (Clustering):

  • Introduction and Definitions
  • Partitioning Methods (K-Means, K-Medoids, FCM, PCM)
  • Model-Based Methods (GMM, EM, MeanShift)
  • Density-Based Methods (DBSCAN)
  • Hierarchical Methods (Agglomerative, Divisive)
  • Spectral Clustering
  • Competitive Learning Algorithms (SOM)
  • Cluster Validity Indices

Unsupervised Learning (Dimension Reduction):

  • Feature Selection
  • Feature Generation
  • Linear Methods (PCA, SVD, FA, LDA)
  • Nonlinear Methods (Autoencoder, LEM, LLE, t-SNE, UMAP, Kernel PCA)

References: