Model of neural visual system with self-organizing cells.

Area of Science:

• Computational Neuroscience
• Artificial Intelligence

Background:

• The visual system's ability to recognize patterns adaptively is crucial for higher animals.
• Feature-detecting cells, like those discovered by Hubel and Wiesel, exhibit plasticity (Blakemore and Cooper).

Purpose of the Study:

• To model an adaptive neural visual system capable of developing pattern recognition capabilities.
• To integrate self-organizing cells with Perceptron-like learning for a comprehensive visual system model.

Main Methods:

• Development of a computational model incorporating "self-organizing cells" to mimic feature-detecting cell plasticity.
• Integration of an eye movement control mechanism to optimize cell usage and self-organization speed.
• Validation through computer simulations and hardware simulator experiments.

Main Results:

• "Self-organizing cells" rapidly develop sensitivity to frequently encountered visual features.
• The model demonstrates efficient pattern classification using a reduced number of feature-detecting cells.
• The eye movement control mechanism accelerates the self-organization process.

Conclusions:

• The proposed model successfully replicates adaptive pattern recognition in a neural visual system.
• The combination of self-organizing cells and Perceptron learning offers an efficient approach to visual system modeling.
• The system's efficiency is enhanced by an integrated eye movement control mechanism.