Transfer Learning for Boosted Relational Dependency Networks Through Genetic Algorithm

Machine learning aims at generalizing from observations to induce models that aid decisions when new observations arrive. However, traditional machine learning methods fail at finding patterns from several objects and their relationships. Statistical relational learning goes a step further to discover patterns from relational domains and deal with data under uncertainty. Most machine learning methods, SRL included, assume the training and test data come from the same distribution. Nonetheless, in several scenarios, this assumption does not hold. Transfer learning aims at acting on scenarios like that, leveraging learned knowledge from a source task to improve the performance in a target task when data is scarce. A costly challenge associated with transfer learning in relational domains is mapping from the source and target background knowledge language. This paper proposes GROOT, a framework that applies genetic algorithm-based solutions to discover the best mapping between the source and target tasks and adapt the transferred model. GROOT relies on a set of relational dependency trees built from the source data as a starting point to build the models for the target data. Over generations, individuals carry a possible mapping. They are submitted to genetic operators that recombine subtrees and revise the initial structure tree, enabling a prune or expansion of the branches. Experimental results conducted on Cora, IMDB, UW-CSE, and NELL datasets show that GROOT reaches results better than the baselines in most cases.