Machine Learning - March 2025

Topics

• Resources

• Resources

  No description yet

  Resources

  There are no resources for this lecture yet

• Introduction to Machine Learning

• Introduction to Machine Learning

  • What is machine learning?;
  • The scientific method;
  • Basic principles: supervised and unsupervised learning, reinforcement learning;
  • Preparing data;
  • Preview of the entire process.

  Resources

  There are no resources for this lecture yet

• Linear and Logistic Regression

• Linear and Logistic Regression

  • Regression – problem statement and motivation;
  • Ordinary least squares: method, simulated example, implementation on real data;
  • RANSAC – robust regression model;
  • Linear regression extensions: polynomial regression, multi-dimensional linear regression;
  • Classification – problem statement and motivation;
  • Logistic regression: method, simulated example, real data;
  • Regularization for regression and classification.

  Resources

  There are no resources for this lecture yet

• Model Training and Improvement

• Model Training and Improvement

  • Training and testing set;
  • Bias-variance tradeoff;
  • k-fold cross-validation;
  • Graphical methods: train/test curve, ROC, confusion matrix;
  • Hyperparameters. Hyperparameter optimization;
  • Model selection.

  Resources

  There are no resources for this lecture yet

• Tree and Ensemble Methods

• Tree and Ensemble Methods

  • Decision trees: information gain and entropy;
  • Random forests;
  • Boosting, AdaBoost;
  • kNN (k nearest neighbors);
  • Applications: regression and classification.

  Resources

  There are no resources for this lecture yet

• Support Vector Machines

• Support Vector Machines

  • SVM intuition: maximum margin;
  • Kernels. "Kernel trick";
  • Applications: regression and classification.

  Resources

  There are no resources for this lecture yet

• Clustering

• Clustering

  • Unsupervised learning: problem statement, intuition, challenges;
  • k-means clustering – motivation, example, kMeans++;
  • Hierarchical clustering – motivation, example;
  • Comparison between k-means and HC: pros and cons;
  • DBSCAN.

  Resources

  There are no resources for this lecture yet

• Dimensionality Reduction

• Dimensionality Reduction

  • Problem statement, intuition;
  • Feature selection vs. feature extraction;
  • Low variance filter, high correlation filter;
  • Random forests as DR algorithms;
  • Principal Component Analysis (PCA): motivation, example, interpretation;
  • Isometric mapping for DR and visualization.

  Resources

  There are no resources for this lecture yet

• Introduction to Neural Networks

• Introduction to Neural Networks

  • Problem statement, motivation and results. Pros and cons;
  • NN representation;
  • Perceptron, softmax function;
  • Two-class and multi-class problems;
  • Feed-forward NNs: forward propagation;
  • Chain rule. Backpropagation and learning;
  • Regularization. Cross-entropy cost function;
  • NN examples with real data;
  • Convolutional NNs: ideas, example

  Resources

  There are no resources for this lecture yet