Understanding Domain-Size Generalization in Markov Logic Networks

MLNs exhibit poor generalization across domain sizes due to parameter variance, which can be minimized for better generalization.