I can't point you to any specific paper/project, but this seems pretty identical to a lot of stuff in genetic algorithm neural network learning. Which was basically discarded as an active area of research after massively parallel gradient descent on GPU's became possible. That said, I do still find this interesting, even if it has been covered before. Memory as an integral part of the process might be kind of novel, but I'm sure someone's tried it before. What is your philosophical goal in constructing this project?
I can't point you to any specific paper/project, but this seems pretty identical to a lot of stuff in genetic algorithm neural network learning. Which was basically discarded as an active area of research after massively parallel gradient descent on GPU's became possible. That said, I do still find this interesting, even if it has been covered before. Memory as an integral part of the process might be kind of novel, but I'm sure someone's tried it before. What is your philosophical goal in constructing this project?