0
$\begingroup$

I am training a CNN model for my specific problem. I have divided the dataset into 70% training set, 20% validation set, and 10% test set. The validation accuracy achieved was 95% and the test accuracy achieved was also 95%. What does this mean? Is this mean that the model is not biased ( not biased to the samples in the validation set ) and its hyperparameters have been fine-tuned correctly? Also, do these results confirm the generalization ability of the model ( no overfitting)?

$\endgroup$

1 Answer 1

0
$\begingroup$

First of all, make sure you did the split before any kind of pre-processing. Splitting data after pre-processing introduces data leakage.

Second, shuffle the data once again, re-train, validate and test, check if the result persists.

If yes, you are right, the model is not biased to the validation set and hyper-parameters have been fine-tuned correctly.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.