December 2017
Intermediate to advanced
536 pages
14h 23m
English
Using ConvNets, we increased our performance on the MNIST dataset reaching almost 95 percent accuracy. Our ConvNet consists of two layers combining convolutions, ReLU, and maxpooling, followed by two fully connected layers with dropout. Training happens in batches of size 128 with Adam used as an optimizer, a learning rate of 0.001, and a maximum number of 500 iterations.
Read now
Unlock full access