Induction of Non-monotonic Logic Programs To Explain Statistical Learning Models

2019/09/18 by Farhad Shakerin · 1 citation
Computer Science · #cs.AI #cs.LG #cs.LO

paper · pdf · doi:10.4204/eptcs.306.51

published as EPTCS 306, 2019, pp. 379-388 · In Proceedings ICLP 2019, arXiv:1909.07646. arXiv admin note: substantial text overlap with arXiv:1808.00629, arXiv:1905.11226, arXiv:1802.06462, arXiv:1707.02693

arxiv created 2019/09/18 · arxiv updated 2019/09/20

Abstract

We present a fast and scalable algorithm to induce non-monotonic logic programs from statistical learning models. We reduce the problem of search for best clauses to instances of the High-Utility Itemset Mining (HUIM) problem. In the HUIM problem, feature values and their importance are treated as transactions and utilities respectively. We make use of TreeExplainer, a fast and scalable implementation of the Explainable AI tool SHAP, to extract locally important features and their weights from ensemble tree models. Our experiments with UCI standard benchmarks suggest a significant improvement in terms of classification evaluation metrics and running time of the training algorithm compared to ALEPH, a state-of-the-art Inductive Logic Programming (ILP) system.

Citations

• Proceedings 35th International Conference on Logic Programming (Technical Communications)
• Induction of Non-Monotonic Logic Programs to Explain Boosted Tree Models Using LIME
• Induction of Non-Monotonic Rules From Statistical Learning Models Using High-Utility Itemset Mining
• Heuristic Based Induction of Answer Set Programs: From Default theories to combinatorial problems
• A New Algorithm to Automate Inductive Learning of Default Theories

Cited by

• FOLD-R++: A Scalable Toolset for Automated Inductive Learning of Default Theories from Mixed Data