bazilas
commented
8 years ago You could create an Error object (dagNN format) where you will implement the mean squared error. Then, you only have to include it to your network definition.
Open kbenahmed89 opened 8 years ago
bazilas
commented
8 years ago You could create an Error object (dagNN format) where you will implement the mean squared error. Then, you only have to include it to your network definition.
Hello,
I'm tuning a pre-trained network using a small dataset (32 MRI images). I need a stoping criterion to know on which epoch I need to stop. I can't use the top 1 error because my training data is small. I want to show the square mean error for each iteration. can you please show me how and what to modify.