Transfer learning when Dropout in source model

by Aadnan Farooq A   Last Updated July 12, 2019 09:19 AM - source

Anyone can you pls help me to understand that. If we have dropout layers in between the convolutional layers of source model. When we do transfer learning from source to target model, the convotional layers where dropout used for source model, how the weights of those layers will be transferred to target layer? The dropout layer randomly pick ā€˜nā€™ neurons during the training so i am confuse how weights will transfer to target model in that case.



Related Questions



Dropout and hyperparameters

Updated May 17, 2017 13:19 PM