afcarl / pygpgo

Bayesian optimization for Python

pyGPGO: Bayesian Optimization for Python

pyGPGO is a simple and modular Python (>3.5) package for bayesian optimization.

Bayesian optimization is a framework that can be used in situations where:

• Your objective function may not have a closed-form. (e.g. the result of a simulation)
• No gradient information is available.
• Function evaluations may be noisy.
• Evaluations are expensive (time/cost-wise)

Installation

Retrieve the latest stable release from pyPI:

pip install pyGPGO

Or if you're feeling adventurous, retrieve it from this repo,

pip install git+https://github.com/hawk31/pyGPGO

Check our documentation in http://pygpgo.readthedocs.io/.

Dependencies

• Typical Python scientific stuff: numpy, scipy.
• joblib, used for parallel computation
• scikit-learn, for other surrogates different from Gaussian Processes.
• pyMC3, for integrated acquisition functions and MCMC inference.
• theano for GPU processing.

All dependencies are automatically taken care for in the requirements file.

Features

• Different surrogate models: Gaussian Processes, Student-t Processes, Random Forests, Gradient Boosting Machines.
• Type II Maximum-Likelihood of covariance function hyperparameters.
• MCMC sampling for full-Bayesian inference of hyperparameters (via pyMC3).
• Integrated acquisition functions

A small example!

The user only has to define a function to maximize and a dictionary specifying input space.

import numpy as np
from pyGPGO.covfunc import matern32
from pyGPGO.acquisition import Acquisition
from pyGPGO.surrogates.GaussianProcess import GaussianProcess
from pyGPGO.GPGO import GPGO


def f(x, y):
    # Franke's function (https://www.mathworks.com/help/curvefit/franke.html)
    one = 0.75 * np.exp(-(9 * x - 2) ** 2 / 4 - (9 * y - 2) ** 2 / 4)
    two = 0.75 * np.exp(-(9 * x + 1) ** 2/ 49 - (9 * y + 1) / 10)
    three = 0.5 * np.exp(-(9 * x - 7) ** 2 / 4 - (9 * y -3) ** 2 / 4)
    four = 0.25 * np.exp(-(9 * x - 4) ** 2 - (9 * y - 7) ** 2)
    return one + two + three - four

cov = matern32()
gp = GaussianProcess(cov)
acq = Acquisition(mode='ExpectedImprovement')
param = {'x': ('cont', [0, 1]),
         'y': ('cont', [0, 1])}

np.random.seed(1337)
gpgo = GPGO(gp, acq, f, param)
gpgo.run(max_iter=10)

Check the tutorials and examples folders for more ideas on how to use the software.