1. Overview of machine learning tasks: classification, regression, prediction and areas of application 2. Decision trees: construction, metrics 3. - 4. Support vector method (SVM): formulation, soft margin formulation, solution, use of kernels, kernel trick, classification into multiple classes 5. Introduction to neural networks: types of networks and their architecture, types of learning 6. - 7. Networks of the multilayer perceptron (MLP) type: perceptron, activation functions and their types, layers and learning, relearning, regularization 8. Feedback learning: use and overview of algorithms with a focus on Q-learning 9. Machine learning meta-algorithms: clustering of weak classifiers, random forests, boosting, AdaBoost 10. Fuzzy set: introduction, properties (domain of values, height, carrier, kernel), cuts, membership 11. Fuzzy sets and their extensions: an overview of set and propositional operations and their properties, fuzzy numbers and fuzzy relations 12. Use of fuzzy logic: fuzzification and defuzzification process, linguistic variables 13. Bayesian statistics: definition of basic terms (distribution, prior, posterior), Bayes theorem and its use 14. Bayesian networks (BN): graph representation, model probability distribution (chain rule), an overview of BN learning algorithms
|
This course is focused on selected parts of Soft Computing, such as fuzzy logic and selected machine learning models (decision trees, support vector machines, neural networks and Bayesian networks). It provides both the theoretical background, as well as a practical introduction to currently used frameworks and libraries (Python). The lectures are focused on theoretical analysis and the domain of possible application. The exercises are focused both on the own design and implementation of algorithms, as well as on the transfer of practical experience in the use of software tools.
|