---
title: "A Soft Version of Predicate Invention Based on Structured Sparsity"
authors:
  - "William Yang Wang"
  - "Kathryn Mazaitis"
  - "William W. Cohen"
doi: ""
url: "https://ijcai.org/proceedings/2015/550"
pdf: "https://ijcai.org/proceedings/2015/0550.pdf"
venue: "Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence"
year: 2015
pages: "3918-3923"
track: "Special Track on Machine Learning"
keywords:
---

# A Soft Version of Predicate Invention Based on Structured Sparsity

William Yang Wang, Kathryn Mazaitis, William W. Cohen

*Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence*, pages 3918-3923.

## Abstract

In predicate invention (PI), new predicates are introduced into a logical theory, usually by rewriting a group of closely-related rules to use a common invented predicate as a subroutine. PI is difficult, since a poorly-chosen invented predicate may lead to error cascades. Here we suggest a soft version of predicate invention: instead of explicitly creating new predicates, we implicitly group closely-related rules by using structured sparsity to regularize their parameters together. We show that soft PI, unlike hard PI, consistently improves over previous strong baselines for structure-learning on two large-scale tasks.