-1

I train a CNN model using convolution layers from Resnet50(I freeze all layer in Resnet50). But the loss won't change during epochs running. I think this call gradient vanishing. I am new to Deep Learning so I want to hear from you guy how can we prevent this problem ?

loss didn't change CNN_structure

I expect some counter method for this problem

Nam Ngo
  • 1
  • 2
  • please show code for creating your model and also code for providing data into your model. – Gerry P Nov 13 '22 at 22:04
  • I don't know how to upload file to stack overflow,here my drive you can view (filename Resnet50_model). https://drive.google.com/drive/folders/1vl2n9U0u1er5uJJ5t34RerhnOdeKTcY_ – Nam Ngo Nov 14 '22 at 12:05
  • All model is ok except Resnet50_model. – Nam Ngo Nov 14 '22 at 12:07

1 Answers1

0

You can try whether to change the algorithm of the gradient or change the weights initialization method.

Eloi
  • 101
  • 10
  • As it’s currently written, your answer is unclear. Please [edit] to add additional details that will help others understand how this addresses the question asked. You can find more information on how to write good answers [in the help center](/help/how-to-answer). – Community Nov 19 '22 at 14:18