COMmon Bayesian Optimization
Clone or download
Tsuyoshi Ueno
Tsuyoshi Ueno fix bugs in search.discrete
update tutorial_interactive_mode
Latest commit 7743856 Apr 13, 2016
Failed to load latest commit information.
combo fix bugs in search.discrete Apr 13, 2016
docs add document Feb 18, 2016
examples/grain_bound fix bugs in search.discrete Apr 13, 2016
.gitignore add .gitignore Jan 26, 2016 modify Feb 29, 2016 create a new package for discrete BO Apr 10, 2016

COMmon Bayesian Optimization Library ( COMBO )

Bayesian optimization has been proven as an effective tool in accelerating scientific discovery. A standard implementation (e.g., scikit-learn), however, can accommodate only small training data. COMBO is highly scalable due to an efficient protocol that employs Thompson sampling, random feature maps, one-rank Cholesky update and automatic hyperparameter tuning. Technical features are described in our document.

Required Packages

  • Python 2.7.x
  • numpy >=1.10
  • scipy >= 0.16
  • Cython >= 0.22.1
  • mpi4py >= 2.0 (optional)


1. Download or clone the github repository, e.g.
	> git clone

2. Run install
	> cd combo
	> python install


1. Delete all installed files, e.g.
	> python install --record file.txt
	> cat file.txt  | xargs rm -rvf


After installation, you can launch the test suite from 'examples/grain_bound/tutorial.ipynb'.


This package is distributed under the MIT License.