Lifted Weight Learning of Markov Logic Networks Revisited

TL;DR

This paper presents a polynomial-time algorithm for maximum-likelihood learning of 2-variable Markov logic networks, leveraging lifted inference and maximum entropy methods to improve efficiency.

Contribution

It introduces a new algorithm for lifted weight learning of 2-variable Markov logic networks with polynomial runtime, based on existing inference and entropy computation techniques.

Findings

Algorithm runs in polynomial time in domain size

Applicable specifically to 2-variable Markov logic networks

Builds on lifted inference and maximum entropy algorithms

Abstract

We study lifted weight learning of Markov logic networks. We show that there is an algorithm for maximum-likelihood learning of 2-variable Markov logic networks which runs in time polynomial in the domain size. Our results are based on existing lifted-inference algorithms and recent algorithmic results on computing maximum entropy distributions.