Structure Learning for Relational Logistic Regression: An Ensemble Approach
This work addresses the challenge of structure learning in relational models for researchers in machine learning and AI, representing an incremental improvement by adapting existing boosting techniques to RLR.
The paper tackles the problem of learning Relational Logistic Regression (RLR) by developing an algorithm based on functional-gradient boosting to learn vector-weighted first-order formulae, demonstrating superiority over other methods in empirical evaluations on standard and novel datasets.
We consider the problem of learning Relational Logistic Regression (RLR). Unlike standard logistic regression, the features of RLRs are first-order formulae with associated weight vectors instead of scalar weights. We turn the problem of learning RLR to learning these vector-weighted formulae and develop a learning algorithm based on the recently successful functional-gradient boosting methods for probabilistic logic models. We derive the functional gradients and show how weights can be learned simultaneously in an efficient manner. Our empirical evaluation on standard and novel data sets demonstrates the superiority of our approach over other methods for learning RLR.