Presentation slides and generated data for Evolutionary Algorithms II about Estimation of Distribution Algorithms.
- The presentation slides should also be present here.
- Sorry fot not showing (the better) half of them.
The notebook contains all the necessary code to recreate the graphs contained in the presentation as well as the source code for a few algorithms. I recommend running it directly in Google Colab:
- Colab (Graphs must be rerun, it does not save them :/ )
- Do not forget to upload already generated
.npydata in order to avoid generating them, it takes a while. - If you do not see any graphs after running the cells, try running them one by one. At least my Opera browser gets very laggy and refuses to show some if I run them all via "Run all" button.
- The sliders are little laggy too as the tutorials from plotly did not work.
- Feel free to point out any mistakes.
I did try to implement a simple VAE-Q version, but it does not work very well, I did not use Tournament selection, maybe that's why? Or I made some mistakes somewhere.
Also did not work very well, gets stuck in local maxima all the time. But the model and the encoding is not very elaborate. Definitely see this presentation https://www.slideshare.net/InformaticaUCM/dealing-with-constraints-in-estimation-of-distribution-algorithms for more information on constrained models.