We host Let's Talk ML, regular informal meetups at Faculty of Information Technology. Meetups are held every two weeks. Two short 20 minute talks are followed by discussion. Topics cover all the hot buzzwords like Machine Learning, Artificial Intelligence, Data Science, Deep Learning, NLP, Rainbows, Kittens, Meta learning, Meta-meta learning, Meta-meta-meta learning... Already spotted a pattern? Join us at the next Let's Talk!
Our scientists directly lead or are involved in teaching several courses at Faculty of Information Technology.
Students are introduced to the fundamental problems of the Artificial Intelligence, and the basic methods for their solving. It focuses mainly on the classical tasks from the areas of state space search, multi-agent systems, game theory, planning, and machine learning. Modern soft-computing methods, including the evolutionary algorithms and the neural networks, will be presented as well.
Students are introduced to the basic methods of discovering knowledge in data. In particular, they learn the basic techniques of data preprocessing, multidimensional data visualization, statistical techniques of data transformation, and fundamental principles of knowledge discovery methods. Students will be aware of the relationships between model bias and variance and will know the fundamentals of assessing model quality. Data mining software is extensively used in the module. Students will be able to apply basic data mining tools to common problems (classification, regression, clustering).
Course is focused on deep understanding of most commonly used data mining algorithms. These are mainly decision trees, perceptron and multilayer perceptron, support vector machines, linear polynomial and logistic regression, SOM, k means etc.
Data pre-processing is a crucial step for successful data mining. Data pre-processing steps can take considerable amount of time very often significantly more then processing itself. However, irrelevant, redundant and noisy information in the data complicates the training phase and usually prevents to achieve the best results. Knowledge of algorithms for data cleaning, normalization, transformation, feature extraction and selection from various data sources is a fundamental part of knowledge engineering.
The module gives an overview of basic methods and techniques of computational intelligence that stem from the classical artificial intelligence. Computational intelligence methods are mostly nature-inspired, parallel by nature, and applicable to many problems in knowledge engineering.
The aim of this module is to bring together and to systematize the knowledge and skills received in specialization-specific knowledge engineering modules, as well as to demonstrate its use on real-world problems.
Pattern Recognition is the prerequisite for modern approaches to artificial intelligence, machine perception, computer graphics, and many other related disciplines, such as date mining, hypermedia, etc. Students will learn elements of pattern recognition, Bayesian decision theory, learning theory, parametric and non-parametric classifiers, support vector machines, classification quality estimations, feature selection, and cluster analysis.