A neural network model as a globally coupled map and applications based on chaos

Area of Science:

• Computational Neuroscience
• Artificial Intelligence
• Network Science

Background:

• Hopfield networks are foundational recurrent neural networks.
• Existing models may lack dynamic adaptability for complex information processing tasks.

Purpose of the Study:

• To propose a modified Hopfield network as a globally coupled map (GCM).
• To introduce a dynamic information processing model based on GCM.
• To demonstrate applications in information retrieval and combinatorial optimization.

Main Methods:

• Modification of a Hopfield network with negative self-feedback.
• Interpretation of information processing via network element maps.
• Development of a dynamic information processing framework.

Main Results:

• Successful implementation of the GCM neural network model.
• Demonstration of the model's capacity for dynamic information processing.
• Validation of the model's applicability to vague keyword searches and the traveling salesman problem (TSP).

Conclusions:

• The proposed GCM offers a novel approach to neural network modeling.
• The dynamic information processing model provides a flexible framework for complex tasks.
• The GCM shows promise for practical applications in search and optimization problems.