← Back to Publications

A Novel Strategy for Hybridizing Subsymbolic and Symbolic Learning and Representation

Authors:

Year: 2011
Venue: AAAI Fall Symposium Series (2011)
Links: paper (PDF)

Summary

Proposes integrating subsymbolic compositional spatiotemporal deep-learning networks (CSDLNs — e.g. HTM, DeSTIN) into a symbolic integrative cognitive architecture via dynamic, adaptive linkage between CSDLN attractors (across sensory, motor, and goal hierarchies) and symbolic representations. It introduces “semantic CSDLNs” that retain CSDLN structure while holding more abstract, symbol-like patterns, concretized as a planned DeSTIN–OpenCog integration.

Relevance to Hyperon

A neural-symbolic hybridization source in the PRIMUS/cognitive-synergy lineage — relevant to the PRIMUS Deep Dive's account of bridging subsymbolic perception with symbolic reasoning.

Key References

• RawData source
• PRIMUS Deep Dive