Tutorial ‐ Unsupervised training

IBody

You have to implement a few methods when subclassing IBody.

// Return the number of inputs the network should have.
GetInputCount() -> std::size_t

// Return the number of outputs the network should have.
GetOutputCount() -> std::size_t

// Return the target fitness you want to reach.
GetMaximumFitness() -> Type::fitness_t 

// Return the current fitness of your model
GetFitness() -> Type::fitness_t

// Return a vector of floats representing the inputs.
// Gets called each time the `Update` method gets called.
ProvideNetworkWithInputs() -> Type::neuron_values_t

// Evaluate the outputs of the network and update your fitness.
Update(Type::neuron_values_t& networkOutputs) -> void

// Return false if you need more iterations in this generation (I.e. should the Update method get called once more)
HasFinishedTask() -> bool

// Reset your fitness and other things you might have stored for the current training iteration.
Reset() -> void