Fun

object Fun

The activation functions a neuron can have.

They can be used in the arguments of neural network's constructor.

Net(List(2, 3, 1), _ => Fun.sigmoid, _ => Random().nextDouble())

Net(List(4, 6, 8, 5, 3), _ => Fun.identity, _ => Random().nextDouble())

Net(List(4, 8, 3), _ => Fun.tanh, _ => Random().nextDouble())

Net(List(2, 1), _ => Fun.leakyReLU, _ => Random().nextDouble())

class Object

trait Matchable

class Any

Value members

Identity is a linear function where the output is equal to the input.

x => x

LeakyReLU gives a small proportion of x if x is negative and x otherwise.

x => if x < 0.0 then 0.01 * x else x

Sigmoid takes any real value as input and outputs values in the range of 0 to 1.

x => 1.0 / (1.0 + Math.exp(-x))

Tanh is similar to Sigmoid, but outputs values in the range of -1 and 1.

x => Math.tanh(x)