flxsosa / DeepHyperNEAT

A public python implementation of the DeepHyperNEAT system for evolving neural networks. Developed by Felix Sosa and Kenneth Stanley. See paper here: https://eplex.cs.ucf.edu/papers/sosa_ugrad_report18.pdf
Apache License 2.0
72 stars 15 forks source link

Add example game #5

Open 0T34 opened 2 years ago

0T34 commented 2 years ago

Hi sir, I'm really interested in your HyperNEAT expansion, I'm trying to better understand the algorithm in general so I wrote a game to compare the speed and learning curve of DeepHyperNEAT.

As I don't know all the details of DeepHyperNEAT I couldn't find the most optimal parameters and I think that with a little more work the learning curve could be greatly improved.

One of the problems I encountered is that when the directories were moved in the commit c1707cf6d247642523e79020e9a976399faa1da6 the examples could no longer be executed because the imports were not found, which I solved with: import sys sys.path.append("..") but, this makes it impossible for us to save the images of the neural network, since we change the execution directory.

I would like to know if by any chance you have Discord, Telegram or some way to contact you, because I am really interested in cleaning and optimizing the code and then rewriting it in C with multithreading, in order to have a version for production.

With nothing more to add, thank you very much for your contribution and I look forward to your response.