Struct revonet::neproblem::SymbolicRegressionProblem

pub struct SymbolicRegressionProblem { /* fields omitted */ }

Problem which is typically used to test GP algorithms. Represents symbolic regression with 1 input and 1 output. There are three variants: * f - 4-th order polynomial. * g - 5-th order polynomial. * h - 6-th order polynomial.

See for details: Luke S. Essentials of metaheuristics.

Methods

impl SymbolicRegressionProblem

fn new(problem_type: char) -> SymbolicRegressionProblem

Create a new problem depending on the problem type: * f - 4-th order polynomial. * g - 5-th order polynomial. * h - 6-th order polynomial.

Arguments:

• problem_type - symbol from set ('f', 'g', 'h') to set the problem type.

fn new_f() -> SymbolicRegressionProblem

Create f-type problem (4-th order polynomial)

fn new_g() -> SymbolicRegressionProblem

Create g-type problem (4-th order polynomial)

fn new_h() -> SymbolicRegressionProblem

Create h-type problem (4-th order polynomial)

Trait Implementations

impl NeuroProblem for SymbolicRegressionProblem

fn get_inputs_num(&self) -> usize

Number of input variables.

fn get_outputs_num(&self) -> usize

Number of output (target) variables.

fn get_default_net(&self) -> MultilayeredNetwork

Returns random network with default number of inputs and outputs and some predefined structure.

fn compute_with_net<T: NeuralNetwork>(&self, nn: &mut T) -> f32

Compute fitness value for the given neural network.