We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Artificial Intelligence

Title: Inductive logic programming at 30: a new introduction

Abstract: Inductive logic programming (ILP) is a form of machine learning. The goal of ILP is to induce a hypothesis (a set of logical rules) that generalises given training examples. In contrast to most forms of machine learning, ILP can learn human-readable hypotheses from small amounts of data. As ILP approaches 30, we provide a new introduction to the field. We introduce the necessary logical notation and the main ILP learning settings. We describe the main building blocks of an ILP system. We compare several ILP systems on several dimensions. We describe in detail four systems (Aleph, TILDE, ASPAL, and Metagol). We document some of the main application areas of ILP. Finally, we summarise the current limitations and outline promising directions for future research.
Comments: work in progress
Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Cite as: arXiv:2008.07912 [cs.AI]
  (or arXiv:2008.07912v3 [cs.AI] for this version)

Submission history

From: Andrew Cropper [view email]
[v1] Tue, 18 Aug 2020 13:09:25 GMT (136kb)
[v2] Fri, 9 Oct 2020 12:52:09 GMT (182kb)
[v3] Tue, 13 Oct 2020 16:35:41 GMT (189kb)

Link back to: arXiv, form interface, contact.