Give the weights and structure of a neural network with a sigmoid output activation and one hidden layer with an ReLU activation, that can represent the exclusive-or function (⊕) of two Booleans, which is true when the inputs have different truth values; see Figure 7.13 (page 293). Assume true is represented as 1, and false as 0. [Hint: Write exclusive-or in terms of other logical operators.

See