Analogical Logic Program Synthesis from Examples
Abstract
The purpose of this paper is to present a theory and an algorithm for analogical logic program synthesis from examples. Given a source program and examples, the task of our algorithm is to find a program which explains the examples correctly and is similar to the source program. Although we can define a notion of similarity in various ways, we consider a class of similarities from the viewpoint of how examples are explained by a program. In a word, two programs are said to be similar if they share a common explanation structure at an abstract level. Using this notion of similarity, we formalize an analogical logic program synthesis and show that our algorithm based on a framework of model inference can identify a desired program.
Cite
Text
Sadohara and Haraguchi. "Analogical Logic Program Synthesis from Examples." European Conference on Machine Learning, 1995. doi:10.1007/3-540-59286-5_61Markdown
[Sadohara and Haraguchi. "Analogical Logic Program Synthesis from Examples." European Conference on Machine Learning, 1995.](https://mlanthology.org/ecmlpkdd/1995/sadohara1995ecml-analogical/) doi:10.1007/3-540-59286-5_61BibTeX
@inproceedings{sadohara1995ecml-analogical,
title = {{Analogical Logic Program Synthesis from Examples}},
author = {Sadohara, Ken and Haraguchi, Makoto},
booktitle = {European Conference on Machine Learning},
year = {1995},
pages = {232-244},
doi = {10.1007/3-540-59286-5_61},
url = {https://mlanthology.org/ecmlpkdd/1995/sadohara1995ecml-analogical/}
}