Bayesian logic networks (BLN) and Markov logic networks (MLN) are two powerful tools of statistical

relational learning (SRL) which is a subdiscipline of artificial intelligence and machine learning. In

general, SRL deals with models of domains that exhibit both uncertainty and relational structure.

Methods developed in SRL are well suited for (but not limited to) knowledge representation.

They use first-order logic to describe relational properties of a domain in a general manner and

adopt probabilistic graphical models (such as Bayesian networks or Markov networks) to model the

uncertainty.

 

In this Bachelor/Master thesis one should define and model common sense knowledge for indoor robot

mapping [1] using either BLN or MLN, so as to enhance its performance. The underlying idea of this

thesis is to demonstrate how the use of high-level knowledge (rule-based) could possibly ease low-level

data processing. The BLN/MLN that is used for knowledge modeling must be trained using labeled data, and the final approach should be experimentally evaluated.