Learning to Approximate: Auto Direction Vector Set Generation for Hypervolume Contribution Approximation

The data is organized as below:

• Folder LtA contains training solution sets and learned direction vector sets for the learning process of LtA.
• Folder VectorSets contains the direction vector sets generated by DAS, UNV, JAS, MSS-D, MSS-U, Kmeans-U.
• Folder CIR_solution_sets contains the solution sets for the Correct Identification Rate experiment.
• Folder Candidate_solution_sets contains the candidate solution sets for the GAHSS experiment.
• Folder LtA-S contains the single traning solution set and the learned direction vector sets based on it.
• Folder LtA-M contains the modified direction vector set based on the learned direction vector sets.
• Folder Code for Figure 1 contains the source code and data for the experiment in Fig. 1.

This paper has been submitted to a journal for review. The source code of LtA will be released if the paper is published or the reviewers request.