Hands-On Q-Learning with Python

ReLU functions

A ReLU function is a type of activation function. As previously discussed, an activation function provides the output of a node in a network (such as 0 or 1). In the case of a ReLU function, our outputs will be a wider range of integer values.

Here is a very simple implementation of a ReLU function:

def ReLU(x):    if x > 0:        return x    return 0

For an input x, the ReLU function returns x if x is positive, and 0 if it is not. The ReLU is unbounded on the right-side and can go to infinity. Effectively, the ReLU is flattening the negative inputs to zero and only keeping the positive inputs.

We won't need to implement an actual ReLU function ourselves in this way; the function will be packaged into a framework such as TensorFlow for ...