Module reinforcement_learning 

Reinforcement Learning Optimization Module

This module implements optimization algorithms based on reinforcement learning principles, where optimization strategies are learned through interaction with the objective function environment.

Key Features

• Policy Gradient Optimization: Learn optimization policies using policy gradients
• Q-Learning for Optimization: Value-based approach to optimization strategy learning
• Actor-Critic Methods: Combined policy and value learning for optimization
• Bandit-based Optimization: Multi-armed bandit approaches for hyperparameter tuning
• Evolutionary Strategies: Population-based RL optimization
• Meta-Learning: Learning to optimize across different problem classes

Applications

• Automatic hyperparameter tuning
• Adaptive optimization algorithms
• Black-box optimization
• Neural architecture search
• AutoML optimization pipelines

Re-exports

pub use actor_critic::*;

pub use bandit_optimization::*;

pub use evolutionary_strategies::*;

pub use meta_learning::*;

pub use policy_gradient::*;

pub use q_learning_optimization::*;

Modules

actor_critic

Actor-Critic Methods for Optimization

bandit_optimization

Multi-Armed Bandit Optimization

evolutionary_strategies

Evolutionary Strategies for RL Optimization

meta_learning

Meta-Learning for Optimization

policy_gradient

Advanced Policy Gradient Optimization with Meta-Gradient Learning

q_learning_optimization

Q-Learning for Optimization

utils

Utility functions for RL optimization

Structs

ConvergenceMetrics

Convergence metrics for RL state

Experience

Experience tuple for RL

ExperienceBuffer

Experience replay buffer

ImprovementReward

Simple improvement-based reward function

OptimizationState

State representation for optimization RL

RLOptimizationConfig

Configuration for reinforcement learning optimization

Enums

OptimizationAction

Action space for optimization RL

Traits

RLOptimizer

Trait for RL-based optimizers

RewardFunction

Reward function for optimization RL