Hands-On Neuroevolution with Python

XOR experiment hyperparameters

The XOR experiment starts with a very straightforward initial genome configuration that has only two input nodes, one output node, and one special input—the bias node. No hidden node is introduced in the initial genome:

[DefaultGenome]# The network parametersnum_hidden = 0num_inputs = 2num_outputs = 1# node bias optionsbias_init_mean = 0.0bias_init_stdev = 1.0

The activation function of all the network nodes is a sigmoid and the node inputs are aggregated by the sum function:

[DefaultGenome]# node activation optionsactivation_default = sigmoid# node aggregation optionsaggregation_default = sum

The type of the encoded network is feed-forward fully connected:

[DefaultGenome]feed_forward = Trueinitial_connection ...