Learning and analyzing vector encoding of symbolic representations
Artificial Intelligence
2018-03-13 v1
Abstract
We present a formal language with expressions denoting general symbol structures and queries which access information in those structures. A sequence-to-sequence network processing this language learns to encode symbol structures and query them. The learned representation (approximately) shares a simple linearity property with theoretical techniques for performing this task.
Cite
@article{arxiv.1803.03834,
title = {Learning and analyzing vector encoding of symbolic representations},
author = {Roland Fernandez and Asli Celikyilmaz and Rishabh Singh and Paul Smolensky},
journal= {arXiv preprint arXiv:1803.03834},
year = {2018}
}