Dynamic control of an artificial neural system: the property inheritance network

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Artificial Intelligence

Background:

• Hierarchical content addressable memory systems are crucial for efficient information retrieval.
• Adaptive Resonance Theory (ART) provides a framework for stable, self-organizing neural networks.
• Property inheritance mechanisms are fundamental to knowledge representation.

Purpose of the Study:

• To introduce and describe the Property Inheritance Network (PIN) architecture.
• To detail the implementation of associative memory using adaptive resonance circuits within the PIN.
• To explain the control mechanisms for sequential information search in a knowledge taxonomy.

Main Methods:

• Review of knowledge representation concepts.
• Summary of Carpenter and Grossberg's Adaptive Resonance Theory.
• Presentation of the PIN architecture and control implementation.
• Discussion of simulation results.

Main Results:

• The PIN architecture effectively utilizes adaptive resonance circuits for associative memory.
• Control neurons successfully manage sequential searches through stored information taxonomies.
• Simulations demonstrate the feasibility and functionality of the PIN system.

Conclusions:

• The Property Inheritance Network offers a novel approach to dynamic information access in hierarchical memory.
• Adaptive resonance circuits are well-suited for implementing the associative memory component of the PIN.
• The control system provides effective state-dependent navigation of stored knowledge.