February 2018
Intermediate to advanced
262 pages
6h 59m
English
After we convert the data from a two-dimensional tensor to a one-dimensional tensor, we pass the data through a linear layer, followed by a nonlinear activation layer. In our architecture, we have two linear layers; one followed by ReLU, and the other followed by a log_softmax, which predicts what digit is contained in the given image.
Read now
Unlock full access