Goal:
The aim of the course is to enable students to transform domain-specific practical problems into machine learning tasks and to critically evaluate the properties of various machine learning algorithms. Students will also learn to select and apply the appropriate methods for a given learning objective. In addition, they will gain knowledge of a wide range of data preprocessing techniques and will be prepared to evaluate machine learning methods in terms of their applicability.
Course description:
Principles of machine learning. Data preprocessing techniques. Unsupervised learning: clustering methods. Supervised learning for predicting nominal class labels: classification. Supervised learning for predicting continuous output variables: regression. Model selection. Feature selection. Fundamental techniques (1R, Naïve Bayes, covering algorithms), Bayesian networks, k-NN, inductive learning and decision trees, support vector machines, Random Forests, ensemble learning models. Reinforcement learning. Evaluation of model performance. Quality and error measures, cross-validation. Bias–variance trade-off. AutoML. Following the theoretical material, students examine and compare various machine learning algorithms on diverse datasets during practical and laboratory sessions, through individual and group project assignments, primarily using Python packages.