Spiking Neural P Systems With Learning Functions.

Area of Science:

• Computational Neuroscience
• Artificial Intelligence

Background:

• Spiking neural P systems (SN P systems) are distributed, parallel computing models inspired by neuronal communication via spikes.
• Existing SN P systems have primarily focused on theoretical advancements.

Purpose of the Study:

• Introduce a novel variant: SN P systems with learning functions, enabling dynamic connection adjustments.
• Evaluate the efficacy of these learning SN P systems in pattern recognition, specifically English letter recognition.

Main Methods:

• Developed specific SN P systems incorporating a simple Hebbian learning function.
• Tested the systems on English letter recognition tasks under varying noise levels (none, low, medium, high).
• Compared performance against backpropagation neural networks, probabilistic neural networks, and standard spiking neural networks.

Main Results:

• Achieved an average accuracy rate of 98.76% in noise-free English letter recognition.
• Demonstrated superior performance over backpropagation and probabilistic neural networks in noisy environments.
• Showed slightly better performance than standard spiking neural networks in recognizing letters with noise.

Conclusions:

• SN P systems with learning functions are effective for pattern recognition tasks.
• This study marks the first application of SN P systems in pattern recognition, validating their practical feasibility.
• The developed systems offer a promising approach for robust pattern recognition, especially in the presence of noise.