3D Multiplex Cross-Frequency Graph Attention Network for Real-Time Closed-Loop Electrophysiology Control

This dataset and research record accompanies the paper titled "3D Multiplex Cross-Frequency Graph Attention Network for Real-Time Closed-Loop Electrophysiology Control," which introduces a hardware-efficient neuro-AI framework for real-time brain-computer interfaces. The repository archives the complete model assets, serialization configurations, and experimental evaluation artifacts—including the 3D multiplex connectivity tensor topology, INT8 quantized ONNX binaries, and latency benchmarking logs—demonstrating sub-millisecond core inference (0.1429 ms) and safety-constrained closed-loop neuromodulation suitable for resource-constrained edge microcontrollers.