Module interactive 

Interactive Genetic Algorithm (IGA) module

This module provides support for human-in-the-loop evolutionary optimization, where fitness is derived from user preferences rather than an automated function.

Overview

Interactive GAs are useful when:

• The fitness function cannot be easily formalized
• Human aesthetic judgment is needed (art, design, music generation)
• User preferences are subjective and vary per individual

Evaluation Modes

The module supports three interaction paradigms:

• Rating: Users assign numeric scores to individual candidates
• Pairwise Comparison: Users pick the better of two candidates
• Batch Selection: Users select their favorites from a presented batch

Example

use fugue_evo::interactive::prelude::*;
use fugue_evo::prelude::*;

let mut iga = InteractiveGABuilder::<RealVector>::new()
    .population_size(12)
    .evaluation_mode(EvaluationMode::BatchSelection)
    .batch_size(6)
    .bounds(bounds)
    .selection(TournamentSelection::new(2))
    .crossover(SbxCrossover::new(15.0))
    .mutation(PolynomialMutation::new(20.0))
    .build()?;

loop {
    match iga.step(&mut rng) {
        StepResult::NeedsEvaluation(request) => {
            let response = present_to_user(&request);
            iga.provide_response(response);
        }
        StepResult::GenerationComplete { generation, .. } => {
            println!("Generation {} complete", generation);
        }
        StepResult::Complete(result) => break,
    }
}

Modules

aggregation

Fitness aggregation models for interactive evaluation

algorithm

Interactive Genetic Algorithm implementation

bradley_terry

Bradley-Terry model implementation with Maximum Likelihood Estimation

evaluator

Core types for interactive evaluation

prelude

Prelude for convenient imports

selection_strategy

Active learning strategies for intelligent candidate selection

session

Session state management for interactive evolution

traits

Interactive fitness traits

uncertainty

Uncertainty quantification for fitness estimates