🧠 NeuroGraph-TSC: A Neuro-Inspired Graph-Based Temporal-Spatial Classifier for Cognitive State Prediction from EEG
Minimal GNN + LSTM pipeline for EEG with 32 electrodes:
- Spatial modeling with Graph Attention Networks (GAT) over an electrode graph
- Temporal modeling with LSTM across time
- Optional neuroscience-inspired regularizers based on neural mass models
Figure 1. Overview of the NeuroGraph-TSC architecture showing spatial graph encoding, temporal sequence modeling, and neural mass model-based regularization.
# 1) Create environment and install dependencies pip install -r requirements.txt # 2) (PyG wheels) — choose your torch/CUDA combo: # See https://data.pyg.org/whl/ for the right URL. Example (Torch 2.2 + CUDA 12.1): pip install pyg-lib torch-scatter torch-sparse torch-cluster torch-spline-conv \ -f https://data.pyg.org/whl/torch-2.2.0+cu121.html pip install torch-geometric # 3) Prepare your numpy arrays: # data.npy shape (N, T, 32) # labels.npy shape (N,) (strings or ints) python scripts/run_train.py \ --data-path /path/to/data.npy \ --labels-path /path/to/labels.npy \ --epochs 20 --time-steps 100