Machine Learning applied to the optimization of the HPX backend of Blaze. This repository allows to generate data using blazemark and allows to fit machine learning models using scikit-learn library and our own custom Decision Tree Classifier.
The repository is structured as follow:
-
Data Generation (contains the bash scripts that are run to generate data files)
-
Data Analysis ( contains python scripts to analyze and vizualize the data generated. Machine learning algorithms are also fit on the Training Set and Evaluated on the Test Set)
-
Benchmarks ( contains python scripts to plot performance graphs for different benchmarks. This allows to compare the old HPX backend and the Machine Learning backend)
-
Models ( contains the header files that represent the fitted classification trees )
