Machine learning is a very popular domain of computer science that allows for learning from data models for classification, prediction or learning a strategy in unknown environments. The model or the strategy can be expressed using many methods. 

Lately, non-symbolic knowledge representation in a form of a (deep) neural network is very popular because it is very effective. However, there are applications in which interpretability of models is more important than a top accuracy. In this presentation examples of such cases together with algorithms using symbolic knowledge representation that can be applied will be discussed.