New answers tagged

1 vote

Splitting medical dataset by patient

There's no whatsoever problem with unbalanced test folds (other than that you need to think how to properly aggregate the results - on scan vs. on patient level - but that's a consequence of the data ...
user avatar
0 votes

few images in validation and test set

If the quantity is equal to the tests then it would be true however if both waiver and quantity the quality would be built-in and no test sets would be required. Professor Bill Maher
user avatar
  • 1
4 votes

few images in validation and test set

Having validation and test set so much small with few samples for defect could be a problem? It, typically, is a problem. With such small numbers, performance evaluations may have large variances. ...
user avatar
  • 52.2k
1 vote

What is the purpose of 2 fully connected hidden layers in VGG16?

Once you've extracted the features from an image, using all those convolutional layers, you need to link the features to the output/target. This is what the final layers do. In simpler neural networks,...
user avatar
  • 5,466

Top 50 recent answers are included