Abstracting Markov Networks

Lorenza Saitta , Christel Vrain (1)

1 LIFO - Laboratoire d'Informatique Fondamentale d'Orléans

In this paper we describe a preliminary investigation on the use of abstraction operators to reduce the complexity of inference in Markov Networks. More specifically, we are interested in Logic Markov Network, where the use of abstraction may be a complementary approach to lifted inference.

Domains: Machine Learning [cs.LG]

Lorenza Saitta, Christel Vrain. Abstracting Markov Networks. WARA-2010, AAAI-10 Workshop on Abstraction, Reformulation, and Approximation, Jul 2010, Atlanta, United States. ⟨hal-00484619⟩