FineTuning

[samane....@gmail.com]

Dear hagaygarty,

First of all, I would like to thank you for your great work.

I have totally 80 mri images with huge size (around [128 480 256]) for 3 classes. If I divide it into test and train, I think the number of data is not enough for training. What should I do? How can I do fine tuning for the network?

Your help is really appreciated,

Regards,

Samane

[Hagay Garty]

One option is to crop smaller patches from each MRI image, you can increase the training set this way.

Another is simply to get more images

[samane....@gmail.com]

What about fine tuning; using the model weight of another trained data, any option in your code?

Regards,

samane 

[Hagay Garty]

Yes, simply call Train again using the 'net' variable returned from previous Train call.