Code Repository for SOLVE

Bentley, P. J., Lim, S. L., Gaier, A. and Tran, L. (2022) Evolving Through the Looking Glass: Learning Improved Search Spaces with Variational Autoencoders. In G. Rudolph et al. (Eds.): Proc of the 17th International Conference on Parallel Problem Solving From Nature (PPSN XVII). Sept 10-14th 2022. Dortmund Germany, LNCS 13398, pp. 371-384, 2022. doi.org/10.1007/978-3-031-14714-2_26

SOLVE extends the original COIL project (Constrained Optimization in Learned Latent Space) to extra criteria in addition to constraints, with layers/stacks and normalized flow VAE. Naming still reflects the original project.

Raw research code used to perform all experiments is in the `v0` branch.

Details of code contents and structure is located in README.txt