Hello, Dr. Barth. i'm the questioner on latest question. i have a question about the sentence about paper.

gher-uliege / DINCAE

DINCAE (Data-Interpolating Convolutional Auto-Encoder) is a neural network to reconstruct missing data in satellite observations.

GNU General Public License v3.0

47 stars 19 forks source link

Hello, Dr. Barth. i'm the questioner on latest question. i have a question about the sentence about paper. #4

Closed jsihunh closed 3 years ago

jsihunh commented 3 years ago

I'm sorry to ask about your paper that doesn't relate to DINCAE. Is there any reason to deduct the time average without cross-validation? As I understand this sentence, cross-validation data will be considered independent data from the total length of the study period. Am I right? or Does any reasons for conduct it?

Kind Regards, Sihun Jung

Alexander-Barth commented 3 years ago

The reason for this is that you do not want to validation data to influence the output of the neural network during training. If the mean is computed over all data, then the mean uses the validation data and the validation data indirectly influences the result during training. I would not expect that the influence is very large but nevertheless it is better to keep the training independent of the validation data.

jsihunh commented 3 years ago

Thank you for your answer.