July 2018
Intermediate to advanced
474 pages
13h 37m
English
Using a deeper CNN with 12 layers (one extra convolution and one extra max pooling layer) resulted in an improvement of accuracy to 99.07%, as shown in the following screenshot:

Using data normalization after every two layers during model building, we were further able to improve the accuracy to 99.85%, as shown in the following screenshot:

You may obtain different results, but feel free to run the training step a few times. The following are some of the steps you can take to experiment with the network in the future to understand ...
Read now
Unlock full access