4 ms·
What I have not seen is the explanation about transferring the data normalization parameters. Say you apply the contrast normalization to the images you use to
by eveningcoffee 10y ago
What I have not seen is the explanation about transferring the data normalization parameters. Say you apply the contrast normalization to the images you use to train the first network.
Is there anything better you can do than applying the same parameters to the second training set?
- argonaut 10y agoIf you had enough data, I imagine you could just re-learn the parameters on the new dataset (and also finetune the network)?