Repository navigation
Expand file tree
/
Copy pathproject_torch.py
More file actions
88 lines (71 loc) · 3.41 KB
/
Copy pathproject_torch.py
File metadata and controls
88 lines (71 loc) · 3.41 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
import torch
from torch import nn, optim
from training import training
from preprocessing import binarize_array, segment
from load_data import load_data_deap
from load_data_amigos import load_data_amigos
from dataset import split_datasets_across_subjects, split_datasets_within_subjects
from models import ModifiedEEGNet
def EEGNet(num_epochs=100, valence=True, arousal=True, deap=True, amigos=True, cross_subject=True, use_scheduler=True):
torch.cuda.empty_cache()
data_deap, labels_deap = load_data_deap()
data_amg, labels_amg = load_data_amigos()
# Preprocessing
labels_deap = binarize_array(labels_deap)
labels_amg = binarize_array(labels_amg)
# Select just valence
if valence and not(arousal):
labels_deap = labels_deap[:, :, 0]
labels_amg = labels_amg[:, :, 0]
num_classes = 1
# Select just arousal
elif not(valence) and arousal:
labels_deap = labels_deap[:, :, 1]
labels_amg = labels_amg[:, :, 1]
num_classes = 1
# Select valence and arousal
else:
labels_deap = labels_deap[:, :, :2]
labels_amg = labels_amg[:, :, :2]
num_classes = 2
# Segment timesteps into 1 second slices
input_length = 128
data_deap, labels_deap = segment(data_deap, labels_deap, input_length)
data_amg, labels_amg = segment(data_amg, labels_amg, input_length)
# Convert to tensors
data_tensor_deap = torch.from_numpy(data_deap)
labels_tensor_deap = torch.from_numpy(labels_deap).to(torch.float32)
data_tensor_amg = torch.from_numpy(data_amg)
labels_tensor_amg = torch.from_numpy(labels_amg).to(torch.float32)
# Creating the sampler
batch_size = 1024
data_1, labels_1, data_2, labels_2 = None, None, None, None
if deap and not(amigos):
data_1 = data_tensor_deap
labels_1 = labels_tensor_deap
elif not(deap) and amigos:
data_1 = data_tensor_amg
labels_1 = labels_tensor_amg
else:
data_1 = data_tensor_deap
labels_1 = labels_tensor_deap
data_2 = data_tensor_amg
labels_2 = labels_tensor_amg
if cross_subject:
train_loader, val_loader, test_loader = split_datasets_across_subjects(data1=data_1, labels1=labels_1, batch_size=batch_size,
data2=data_2, labels2=labels_2)
else:
train_loader, val_loader, test_loader = split_datasets_within_subjects(data1=data_1, labels1=labels_1,
data2=data_2, labels2=labels_2)
# Check for CUDA availability
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print(f"Using device: {device}")
# Initialize the model, loss function, and optimizer
model = ModifiedEEGNet(num_classes=num_classes, input_size=input_length).to(device)
criterion = nn.BCEWithLogitsLoss()
optimizer = optim.Adam(model.parameters(), lr=0.001)
scheduler = None
if use_scheduler:
scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode='min', factor=0.1, patience=20)
training(num_epochs=num_epochs, model=model, train_loaders=train_loader, val_loaders=val_loader, test_loaders=test_loader, criterion=criterion,
optimizer=optimizer, device=device, scheduler=scheduler, num_classes=num_classes)