A General DNA-Like Hybrid Symbiosis Framework: An EEG Cognitive Recognition Method

Area of Science:

• Neuroscience
• Artificial Intelligence
• Computational Biology

Background:

• Artificial neural networks (ANNs) and spiking neural networks (SNNs) are crucial for electroencephalogram (EEG) cognitive recognition.
• Existing research often relies on unidirectional interactions, limiting model performance and creating dependency issues.

Purpose of the Study:

• To propose a novel DNA-like Hybrid Symbiosis (DNA-HS) framework for enhanced EEG cognitive recognition.
• To enable mutual learning between ANNs and SNNs through a bidirectional interaction mechanism.

Main Methods:

• Developed a DNA-like Hybrid Symbiosis (DNA-HS) framework inspired by natural symbiosis.
• Implemented a parametric genetic algorithm and bidirectional interaction for mutual learning between ANNs and SNNs.
• Evaluated the framework by constructing seven hybrid network models for various EEG cognitive recognition tasks.

Main Results:

• The DNA-HS framework significantly improved performance across all tested EEG cognitive recognition tasks.
• Compared to seven traditional models, all hybrid networks using the DNA-HS method showed performance enhancements.
• The proposed method demonstrates effectiveness in improving EEG-based cognitive recognition.

Conclusions:

• The DNA-HS framework offers a new, unified approach for EEG cognitive recognition.
• This symbiotic, DNA-like network structure is expected to establish a new research paradigm.
• The bidirectional learning mechanism enhances model optimization and overall recognition capabilities.