Inductive Logic Boosting
This work addresses a bottleneck in probabilistic logic programming and statistical relational learning for researchers in AI and machine learning, offering an incremental improvement over existing methods.
The paper tackles the challenge of integrating inductive logic programming with statistical learning to estimate probabilities of logic clauses from grounded atoms, proposing an Inductive Logic Boosting framework that transforms relational data into feature-based datasets and relaxes independence constraints. Experimental results show that the learned rules achieve higher AUC-PR and AUC-ROC values than current state-of-the-art SRL methods on benchmark datasets.
Recent years have seen a surge of interest in Probabilistic Logic Programming (PLP) and Statistical Relational Learning (SRL) models that combine logic with probabilities. Structure learning of these systems is an intersection area of Inductive Logic Programming (ILP) and statistical learning (SL). However, ILP cannot deal with probabilities, SL cannot model relational hypothesis. The biggest challenge of integrating these two machine learning frameworks is how to estimate the probability of a logic clause only from the observation of grounded logic atoms. Many current methods models a joint probability by representing clause as graphical model and literals as vertices in it. This model is still too complicate and only can be approximate by pseudo-likelihood. We propose Inductive Logic Boosting framework to transform the relational dataset into a feature-based dataset, induces logic rules by boosting Problog Rule Trees and relaxes the independence constraint of pseudo-likelihood. Experimental evaluation on benchmark datasets demonstrates that the AUC-PR and AUC-ROC value of ILP learned rules are higher than current state-of-the-art SRL methods.