Bayesian Optimization

An extended implementation of Bayesian Optimization.

This is a forked project based on fmfn/BayesianOptimization v1.2.0. Most of the usage and features from the original repository will be kept for a long time.

Installation

You can simply install it with pip command line from the official PyPI site.

pip install hbayes

For more information about installation, you can refer to Installation.

Documentation

The detailed documentation are hosted on https://hansbug.github.io/hbayes/main/index.html.

Only english version is provided now, the chinese documentation is still under development.

Quick Start

A painless example

from hbayes import BayesianOptimization


def black_box_function(x, y):
    """Function with unknown internals we wish to maximize.

    This is just serving as an example, for all intents and
    purposes think of the internals of this function, i.e.: the process
    which generates its output values, as unknown.
    """
    return -x ** 2 - (y - 1) ** 2 + 1


# Bounded region of parameter space
pbounds = {'x': (2, 4), 'y': (-3, 3)}

optimizer = BayesianOptimization(
    f=black_box_function,
    pbounds=pbounds,
    random_state=1,
    verbose=2,
)

optimizer.maximize(
    init_points=10,
    n_iter=25,
)

print(optimizer.max)

The output should be

|   iter    |  target   |     x     |     y     |
-------------------------------------------------
|  1        | -7.135    |  2.834    |  1.322    |
|  2        | -7.78     |  2.0      | -1.186    |
|  3        | -16.13    |  2.294    | -2.446    |
|  4        | -8.341    |  2.373    | -0.9266   |
|  5        | -7.392    |  2.794    |  0.2329   |
|  6        | -7.069    |  2.838    |  1.111    |
|  7        | -6.412    |  2.409    |  2.269    |
|  8        | -3.223    |  2.055    |  1.023    |
|  9        | -7.455    |  2.835    |  0.3521   |
|  10       | -12.11    |  2.281    | -1.811    |
|  11       | -7.0      |  2.0      |  3.0      |
|  12       | -19.0     |  4.0      |  3.0      |
|  13       | -3.383    |  2.0      |  0.3812   |
|  14       | -3.43     |  2.0      |  1.656    |
|  15       | -3.035    |  2.0      |  0.8129   |
|  16       | -17.03    |  4.0      | -0.4244   |
|  17       | -3.012    |  2.0      |  1.109    |
|  18       | -3.0      |  2.0      |  0.9813   |
|  19       | -3.0      |  2.0      |  0.9911   |
|  20       | -3.0      |  2.0      |  0.994    |
|  21       | -3.0      |  2.0      |  0.9957   |
|  22       | -3.0      |  2.0      |  0.9971   |
|  23       | -3.0      |  2.0      |  0.9994   |
|  24       | -3.0      |  2.0      |  1.004    |
|  25       | -3.0      |  2.0      |  0.978    |
|  26       | -3.001    |  2.0      |  1.024    |
|  27       | -3.001    |  2.0      |  0.9735   |
|  28       | -3.001    |  2.0      |  1.024    |
|  29       | -3.001    |  2.0      |  0.9729   |
|  30       | -3.001    |  2.0      |  1.024    |
|  31       | -3.0      |  2.0      |  1.021    |
|  32       | -3.001    |  2.0      |  0.9709   |
|  33       | -3.001    |  2.0      |  0.9749   |
|  34       | -3.001    |  2.0      |  1.023    |
|  35       | -3.001    |  2.0      |  0.9755   |
=================================================
{'target': -3.00000039014846, 'params': {'x': 2.0, 'y': 0.9993753813483197}}

For more tutorial of usages and practices, take a look at Best Practice in documentation.

Contributing

We appreciate all contributions to improve hbayes, both logic and system designs. Please refer to CONTRIBUTING.md for more guides.

License

hbayes released under the MIT license.