An introduction to Markov logic networks and application in video activity analysis

A Markov logic network (MLN) is a compact combination of logic representation of knowledge and probabilistic reasoning in Markov networks. We have seen its applications in different domains, however, few tried to explain or demonstrate the underneath reasons why MLN works. This paper gives an introduction to MLN using examples in the hope to help understand its elegance and booster the application. Application in video activity analysis was designed to demonstrate how MLN can be used in a specific domain, including feature extraction, logic predicate/formula design, and activity recognition through probabilistic reasoning.