Training neural networks to encode symbols enables combinatorial generalization

Author:

Vankov Ivan I.1ORCID,Bowers Jeffrey S.2

Affiliation:

1. Department of Cognitive Science and Psychology, New Bulgarian University, Sofia, Bulgaria

2. School of Psychological Science, University of Bristol, Bristol, UK

Abstract

Combinatorial generalization—the ability to understand and produce novel combinations of already familiar elements—is considered to be a core capacity of the human mind and a major challenge to neural network models. A significant body of research suggests that conventional neural networks cannot solve this problem unless they are endowed with mechanisms specifically engineered for the purpose of representing symbols. In this paper, we introduce a novel way of representing symbolic structures in connectionist terms—the vectors approach to representing symbols (VARS), which allows training standard neural architectures to encode symbolic knowledge explicitly at their output layers. In two simulations, we show that neural networks not only can learn to produce VARS representations, but in doing so they achieve combinatorial generalization in their symbolic and non-symbolic output. This adds to other recent work that has shown improved combinatorial generalization under some training conditions, and raises the question of whether specific mechanisms or training routines are needed to support symbolic processing. This article is part of the theme issue ‘Towards mechanistic models of meaning composition’.

Funder

H2020 European Research Council

European Office for Aerospace Research and Development

Publisher

The Royal Society

Subject

General Agricultural and Biological Sciences,General Biochemistry, Genetics and Molecular Biology

Reference43 articles.

1. Deep learning

2. Mastering the game of Go with deep neural networks and tree search

3. Graves A Mohamed A-R Hinton G. 2013 Speech recognition with deep recurrent neural networks. In Proc. of IEEE Int. Conf. on Acoustics Speech and Signal Processing May 2013 Vancouver British Columbia pp. 6645-6649. Piscataway NJ: IEEE. (doi:10.1109/ICASSP.2013.6638947)

4. Wu Y et al. 2016 Google's neural machine translation system: bridging the gap between human and machine translation. arXiv 1609.08144. (http://arxiv.org/abs/1609.08144)

5. Human-level control through deep reinforcement learning

Cited by 10 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Towards Generalist Biomedical AI;NEJM AI;2024-02-22

2. Human-like scene interpretation by a guided counterstream processing;Proceedings of the National Academy of Sciences;2023-09-28

3. Chapter 10. Connectionist Neuroarchitectures in Cognition and Consciousness Theory Based on Integrative (Synchronization) Mechanisms;Frontiers in Artificial Intelligence and Applications;2023-07-21

4. Configural relations in humans and deep convolutional neural networks;Frontiers in Artificial Intelligence;2023-03-01

5. Modularity in Deep Learning: A Survey;Lecture Notes in Networks and Systems;2023

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3