Symbolické metódy strojového učenia a heuristické vyhľadávanie hypotéz

Abstract

This bachelor thesis addresses the topic of conceptual learning from structured examples represented in first-order predicate logic (PL1). The theoretical part describes the principles of knowledge rep- resentation and incremental learning, focusing on Winston’s algorithm. The practical part presents the design and implementation of a web application that serves as an interactive tool for demon- strating and analyzing this process. The application allows users to upload PL1 datasets, run in- cremental learning, track hypothesis evolution, visualize learned concepts using semantic networks, and compare different hypotheses. The application’s backend is implemented in Python using the FastAPI framework, and the frontend is built with React and TypeScript. The key contribution of the thesis is the creation of a didactic and experimental platform for a better understanding of symbolic machine learning methods and the visualization of complex hypotheses.