MetaGen: A Framework for Metaheuristic Development and Hyperparameter Optimization

🚀 Why MetaGen?

MetaGen simplifies the development of metaheuristics and the optimization of hyperparameters in machine learning and deep learning. Whether you're a researcher, developer, or practitioner, MetaGen provides a structured, flexible, and scalable framework.

🔹 Key Features

✔ Metaheuristic Development Framework – Simplifies the creation of custom metaheuristics.
✔ Hyperparameter Optimization Tools – Supports both layer and architecture-level tuning in deep learning.
✔ Standardized Interface – Ensures compatibility between metaheuristic developers and end users.
✔ Dynamic Architecture Optimization – Adapts model structures dynamically during execution.
✔ Seamless Integration – Compatible with scikit-learn, tensorflow, pytorch, and other ML libraries.
✔ Built-in Metaheuristics – Pre-implemented algorithms ready to use.
✔ Performance Analysis Tools – Integrated visualization tools for analyzing metaheuristic performance.
✔ Scalable and Distributed Execution – Metaheuristics can run efficiently across multiple nodes.

📌 Built-in Metaheuristics

MetaGen includes a collection of powerful metaheuristics:

Random Search
Tabu Search
Simulated Annealing
Tree-Parzen Estimator
Memetic Algorithm
Genetic Algorithm
Steady-State Genetic Algorithm
Coronavirus Optimization Algorithm (CVOA)

📦 Installation

MetaGen requires Python 3.10+ and can be installed with:

pip install pymetagen-datalabupo

📖 Documentation

The official API reference and usage guides are available at: MetaGen Documentation

🛠 Example: Developing a Metaheuristic

Creating a simple Random Search metaheuristic:

from copy import deepcopy
from typing import Callable, List
from metagen.framework import Domain, Solution

class RandomSearch:

    def __init__(self, domain: Domain, fitness: Callable[[Solution], float], search_space_size: int = 30,
                iterations: int = 20) -> None:

        self.domain = domain
        self.fitness = fitness
        self.search_space_size = search_space_size
        self.iterations = iterations

    def run(self) -> Solution:

        potential_solutions: List[Solution] = list()

        for _ in range(0, self.search_space_size):
            potential_solutions.append(Solution(self.domain, connector=self.domain.get_connector()))

        solution: Solution = deepcopy(min(potential_solutions))

        for _ in range(0, self.iterations):
            for ps in potential_solutions:
                ps.mutate()

                ps.evaluate(self.fitness)
                if ps < solution:
                    solution = deepcopy(ps)

        return solution

🤖 Example: Hyperparameter Optimization

Optimizing hyperparameters for a regression model:

from metagen.framework import Domain, Solution
from metagen.metaheuristics.rs import RandomSearch
from sklearn.datasets import make_regression
from sklearn.linear_model import SGDRegressor
from sklearn.model_selection import cross_val_score

# Generate synthetic dataset
X, y = make_regression(n_samples=1000, n_features=4)

# Define the search space
regression_domain = Domain()
regression_domain.define_real("alpha", 0.0001, 0.001)
regression_domain.define_integer("iterations", 5, 200)
regression_domain.define_categorical("loss", ["squared_error", "huber", "epsilon_insensitive"])

# Fitness function
def regression_fitness(solution: Solution):
    model = SGDRegressor(
        loss=solution["loss"],
        alpha=solution["alpha"],
        max_iter=solution["iterations"]
    )
    mape = cross_val_score(model, X, y, scoring="neg_mean_absolute_percentage_error").mean() * -1
    return mape

# Run optimization
best_solution = RandomSearch(regression_domain, regression_fitness).run()
print(best_solution)

🤝 Contributing

We welcome contributions from developers of all experience levels! To contribute:

Open an issue or submit a pull request.
Run tests using:
```
pytest test
```

📌 Resources

MetaGen is an open-source project developed and maintained by:

David Gutiérrez-Avilés
Manuel Jesús Jiménez-Navarro
Francisco José Torres-Maldonado
Francisco Martínez-Álvarez

All authors are members of DataLabUPO, the Data Science & Big Data Research Lab at Pablo de Olavide University.

MetaGen

Install / Use

README