About training the vgg-vd-16 model from scratch

vedaldi / practical-cnn

A VGG practical on convolutional neural networks

154 stars 75 forks source link

Open qiulesun opened 7 years ago

qiulesun commented 7 years ago

I want to train the vgg-vd-16 model from scratch, which parameter initialization method should I choose, gaussian or xavier or xavierimproved?

qiulesun commented 7 years ago

I use the batch-normalization,but the top1err decreased only from 0.99 to 0.91 after 5 epochs, and the convergence rate was slow.Is this normal?