CTLearn Optimizer [GSOC 2019]

Framework for optimizing CTLearn v0.3.0 models developed during the Google Summer of Code 2019.

This optimization utility uses Tune, a scalable framework for hyperparameter search and model training, and supports:

• Random search based optimization.
• Tree Parzen Estimators based optimization.
• Gaussian Processes based optimization.
• Genetic Algorithm based optimization.
• Parallel optimization (depending on available hardware resources).

Authors

• Juan Alfonso Redondo Pizarro

Project dependencies

• CTLearn
• environment_kernels
• Hyperopt
• Matplotlib
• NumPy
• Pandas
• pip
• PyYAML
• Ray
• Scikit-Optimize
• SciPy
• Scikit-learn
• Seaborn
• setproctitle

Documentation dependencies (optional)

The packages listed below are only necessary if you want to build the documentation from the source.

• ipython
• nbsphinx
• pip
• Sphinx
• sphinx-autoapi
• sphinx_rtd_theme

Installation, basic usage and configuration

Take a look at the the documentation.