Structure Learning for Relational Logistic Regression: An Ensemble Approach
- 0 Collaborators
We develop a gradient-boosting technique for learning RLR models. We derive the gradients for the different weights of RLR and show how the rules of the logistic function are learned simultaneously with their corresponding weights. ...learn more
Project status: Concept
Overview / Usage
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. Each clause can be seen as a relational feature for the logistic function. We also note that RLR can be viewed as a probabilistic combination function in that it can stochastically
combine the distributions due to a different set of parents (in graphical model terminology).
Technologies Used
Java