Artificial Intelligence By Example

Flattening

The flattening layer takes the output of the max pooling layer and transforms into a vector of size x * y * z in the following code:

# Step 4 - Flatteningprint("Step 4 Flattening")classifier.add(Flatten())

In this case, the layer vector will be 14 x 14 x 32 = 6,272, as shown in the following diagram:

Flattening layer

This operation creates a standard layer with 6,272 very practical connections for the dense operations that follow. After flattening has been carried out, a fully connected, dense network can be implemented.

Flattening ...