Symbolic Machine Learning (B4M46SMU and BE4M46SMU)

The course will explain methods through which an intelligent agent can learn, that is, improve its behavior by interacting with the environment. The learning scenarios will include

• Concept learning: we will study online learning and batch learning from i.i.d. data. We will define the mistake-bound and PAC model of learning. Strong emphasis will be on logical representations of learned knowledge, including operators for generalization of logic clauses.
• Learning probability distributions with a graphical model (Bayes Networks)
• Reinforcement learning
• Universal learning with the Kolmogorov prior.

Time permitting, we will also discuss active learning with queries. The lectures are given in English for all students.

All lectures and tutorials are held in the regular timetable slots in this MS Teams space.

165 hours ~ 6 ECTS credits