Industrial Genetic Algorithms

[danlovy]

It's been interesting to watch neural nets and 'deep learning' suddenly becoming industrial strength. Companies are investing heavily and there is an arms race to create world domination platforms (AWS, Google etc.).

Genetic algorithms have not crossed that boundary. Any quick YouTube search and you will see a number of very impressive science experiments (the Mario demo in particular).

Is there enough critical mass to leap into an industrial phase? I think so. Here what I think is needed:

1. Set of standards to define how components are built and interact.

2. The ability to create projects that can be built other's work. Right now all the science experiments are one-offs.

3. A published and managed platform. This can be online or local executable. However, the output (like artificial neural net) needs to have a uniform and agreed upon data structure.

Here is some early work on the structure
https://www.youtube.com/watch?v=BaAqFHr0nts


Thoughts?
Dan