Title

Localized Self-Contained Adaptive Networks for Hybrid-Symbolic Reasoning

Abstract

Hybrid-Symbolic processing has been gaining interest over the past decade. This is due to the problems of symbolic representations which are ambiguous, brittle, lack of learning capabilities, and have low availability of parallelism. Sub-symbolic representations have problems of lacking variable binding, symbolic composition and decomposition, and structured representations. Integration of these two representations can mitigate each other's shortcomings. The proposed paradigm: Localized Self-Contained Adaptive Networks (LSCAN) is a localist network using AND and OR evaluators to represent relations between knowledge entities. For optimization of each sub-network, the LSCAN provides learning capabilities for both of feed-forward and lateral relations between network nodes.

Publication Date

12-1-1998

Publication Title

Proceedings of the Joint Conference on Information Sciences

Volume

3

Number of Pages

81-86

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

Socpus ID

1642352837 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/1642352837

This document is currently not available here.

Share

COinS