Stop tracking source directory

source/eeg_motor_imagery.py

@@ -1,142 +0,0 @@
-"""
-EEG Motor Imagery Classification with Shallow ConvNet
------------------------------------------------------
-This script trains and evaluates a ShallowFBCSPNet model
-on motor imagery EEG data stored in .mat files.
-"""
-
-# === 1. Imports ===
-import scipy.io
-import numpy as np
-import mne
-import torch
-from torch.utils.data import TensorDataset, DataLoader
-from sklearn.model_selection import train_test_split
-from braindecode.models import ShallowFBCSPNet
-import torch.nn as nn
-import pandas as pd
-
-# === 2. Data Loading and Epoching ===
-# Load .mat EEG files, create Raw objects, extract events, and epoch the data
-files = [
-    "../data/raw_mat/HaLTSubjectA1602236StLRHandLegTongue.mat",
-    "../data/raw_mat/HaLTSubjectA1603086StLRHandLegTongue.mat",
-    
-]
-
-all_epochs = []
-for f in files:
-    mat = scipy.io.loadmat(f)
-    content = mat['o'][0, 0]
-
-    labels = content[4].flatten()
-    signals = content[5]
-    chan_names_raw = content[6]
-    channels = [ch[0][0] for ch in chan_names_raw]
-    fs = int(content[2][0, 0])
-
-    df = pd.DataFrame(signals, columns=channels).drop(columns=["X5"], errors="ignore")
-    eeg = df.values.T
-    ch_names = df.columns.tolist()
-
-    info = mne.create_info(ch_names=ch_names, sfreq=fs, ch_types="eeg")
-    raw = mne.io.RawArray(eeg, info)
-
-    # Create events
-    onsets = np.where((labels[1:] != 0) & (labels[:-1] == 0))[0] + 1
-    event_codes = labels[onsets].astype(int)
-    events = np.c_[onsets, np.zeros_like(onsets), event_codes]
-
-    # Keep only relevant events
-    mask = np.isin(events[:, 2], np.arange(1, 7))
-    events = events[mask]
-
-    event_id = {
-        "left_hand": 1,
-        "right_hand": 2,
-        "neutral": 3,
-        "left_leg": 4,
-        "tongue": 5,
-        "right_leg": 6,
-    }
-
-    # Epoching
-    epochs = mne.Epochs(
-        raw,
-        events=events,
-        event_id=event_id,
-        tmin=0,
-        tmax=1.5,
-        baseline=None,
-        preload=True,
-    )
-
-    all_epochs.append(epochs)
-
-epochs_all = mne.concatenate_epochs(all_epochs)
-
-# === 3. Minimal Preprocessing + Train/Validation Split ===
-# Convert epochs to numpy arrays (N, C, T) and split into train/val sets
-X = epochs_all.get_data().astype("float32")
-y = (epochs_all.events[:, -1] - 1).astype("int64")  # classes 0..5
-
-X_train, X_val, y_train, y_val = train_test_split(
-    X, y, test_size=0.2, random_state=42, stratify=y
-)
-
-# === 4. Torch DataLoaders ===
-train_ds = TensorDataset(torch.from_numpy(X_train), torch.from_numpy(y_train))
-val_ds   = TensorDataset(torch.from_numpy(X_val),   torch.from_numpy(y_val))
-
-train_loader = DataLoader(train_ds, batch_size=32, shuffle=True)
-val_loader   = DataLoader(val_ds, batch_size=32)
-
-# === 5. Model – Shallow ConvNet ===
-# Reference: Schirrmeister et al. (2017)
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-
-model = ShallowFBCSPNet(
-    n_chans=X.shape[1],
-    n_outputs=len(np.unique(y)),
-    n_times=X.shape[2],
-    final_conv_length="auto"
-).to(device)
-
-# Load pretrained weights
-state_dict = torch.load("model.pth", map_location=device)
-model.load_state_dict(state_dict)
-
-# === 6. Training ===
-criterion = nn.CrossEntropyLoss()
-optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)
-
-for epoch in range(1, 21):
-    # Training
-    model.train()
-    correct, total = 0, 0
-    for xb, yb in train_loader:
-        xb, yb = xb.to(device), yb.to(device)
-        optimizer.zero_grad()
-        out = model(xb)
-        loss = criterion(out, yb)
-        loss.backward()
-        optimizer.step()
-
-        pred = out.argmax(dim=1)
-        correct += (pred == yb).sum().item()
-        total += yb.size(0)
-    train_acc = correct / total
-
-    # Validation
-    model.eval()
-    correct, total = 0, 0
-    with torch.no_grad():
-        for xb, yb in val_loader:
-            xb, yb = xb.to(device), yb.to(device)
-            out = model(xb)
-            pred = out.argmax(dim=1)
-            correct += (pred == yb).sum().item()
-            total += yb.size(0)
-    val_acc = correct / total
-
-    print(f"Epoch {epoch:02d} | Train acc: {train_acc:.3f} | Val acc: {val_acc:.3f}")