Keras Deep Learning Cookbook

Boundary seeking loss

Boundary seeking loss, as we explained in the section at the start of this recipe, is implemented with the following equation:

Where D(x) is the probability that x came from data rather than p_g.

Boundary seeking loss: See the following reference: https://wiseodd.github.io/techblog/2017/03/07/boundary-seeking-gan/.

This is the only major change compared to the previous sample. It is implemented in our class with the following function:

import keras.backend as Kdef boundary_loss(self, y_true, y_pred): return 0.5 * K.mean((K.log(y_pred) - K.log(1 - y_pred))**2)

Next, we will look at how to train this network with the training ...