HOW IMPORTANT ARE NETWORK WEIGHTS? TO WHAT EXTENT DO THEY NEED AN UPDATE?

no code implementations • ICLR 2020 • Fawaz Sammani, Mahmoud Elsayed, Abdelsalam Hamdi

We wish to show that starting from the third epoch, freezing weights which have no informative gradient and are less likely to be changed during training, results in a very slight drop in the overall accuracy (and in sometimes better).

Image Captioning