2

I have used train_test_split function to divide my data into X_train, X_test, y_train, y_test, and then used utils.data.DataLoader to feed it to my CNN but the problem is that I do not know how to access my labels tensor for making a confusion matrix and comparing them with my prediction tensor. I know its a basic question but anyway your help is appreciated.

X_train, X_test, y_train, y_test = train_test_split(faces, emotions, test_size=0.1, random_state=42)
X_train, X_val, y_train, y_val = train_test_split(X_train, y_train, test_size=0.1, random_state=41)

and I used 

train = torch.utils.data.TensorDataset(torch.from_numpy(X_train), torch.from_numpy(y_train))
train_loader = torch.utils.data.DataLoader(train, batch_size=100, shuffle=True)

for feeding the data to my network It seems you can access your labels by just typing targets attribute after your train_set like train_set.targets but it does not work for me that way. How can I get my labels?

Masudur Rahman
  • 1,585
  • 8
  • 16
Hamed Fannan
  • 31
  • 1

1 Answers1

0

PyTorch's DataLoader object is roughly used like this:

for i, (inputs, labels) in enumerate(dataloader):
            inputs = inputs.to(device)
            labels = labels.to(device)

            outputs = model(inputs)
            _, preds = torch.max(outputs, 1)

In general I would suggest to use two DataLoaders, one for training and one for testing/validating. Since you want to make a confusion matrix, you can access your labels simply by your numpy array y_train and your prediction preds e.g. by concatenating them inside the loop to a numpy array.

For more information on how to use the DataLoader, I suggest looking at this very good tutorial: https://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html#sphx-glr-beginner-blitz-cifar10-tutorial-py

and

https://pytorch.org/tutorials/beginner/data_loading_tutorial.html

Tinu
  • 2,432
  • 2
  • 8
  • 20