Module multi_objective_optimization 

Multi-Objective Optimization for Multi-Output Learning

This module provides advanced multi-objective optimization techniques for multi-output learning problems, where multiple conflicting objectives need to be optimized simultaneously. It implements genetic algorithm-based approaches to find Pareto-optimal solutions.

Key Features

• Genetic Algorithm: Population-based evolutionary optimization
• Pareto Optimization: Find trade-off solutions between conflicting objectives
• Non-dominated Sorting: Efficient ranking of solutions using NSGA-II principles
• Crowding Distance: Maintain diversity in the Pareto front
• Multiple Objectives: Support for accuracy, complexity, MSE, MAE, and custom objectives
• Tournament Selection: Efficient parent selection for reproduction

Structs

MultiObjectiveConfig

Configuration for Multi-Objective Optimization

MultiObjectiveOptimizer

Multi-Objective Optimization for multi-output learning

MultiObjectiveOptimizerTrained

Trained state for Multi-Objective Optimizer

ParetoSolution

Pareto-optimal solution