Trajectory-Based Algorithm Selection with Warmstarting - Reproducibility
Diederick Vermetten, Ana Kostovska, Carola Doerr, Tome Eftimov, Jacob de Nobel, Anja Jankovič · Figshare · 2022
The files included accompany the paper "Trajectory-Based Algorithm Selection with Warmstarting" The python-script 'data_collection.py' makes use of the 'dynas.zip' code to run data collection by using IOHexperimenter. The corresponding data is available in 'Performance_Data.zip' (and the collected state features in 'state_features.zip'). The code for generating the figures shown in the paper is included in 'Visualization;ipynb' and 'plot_convergence_curves.R'. The 'data_to_model.zip' contains the results of running the regression models on the performance data, which is then used to create the overall analysis of the models. The 'AlgoSelector.zip' contains all files related to the creation of these models. The remaining pdf-files correspond to figures shown in the paper and their equivalents for other settings which could not fit within the page limit.