3 ms·
The article is from last year but it's still extremely valuable and interesting. Exploring this topic is currently my primary hobby. Specifically, I've been us
by wholemoley 8y ago
The article is from last year but it's still extremely valuable and interesting.
Exploring this topic is currently my primary hobby. Specifically, I've been using OpenAI's retro (Sonic, Contra, Mario, Donkey Kong and, more recently FZero) and comparing the ancient NEAT with more fashionable stuff like DQN, PPO, A3C and DDPG.
With my extremely limited experience, NEAT seems to outperform all of these other algorithms. I believe the advantage is the potential for strange/novel network structure.
And the best part is that NEAT doesn't require a powerful GPU.
Apologies for the shameless plug but here's a link to a series on youtube I made about using Retro and NEAT together to play Sonic. https://www.youtube.com/watch?v=pClGmU1JEsM&list=PLTWFMbPFsvz3CeozHfeuJIXWAJMkPtAdS https://www.youtube.com/watch?v=pClGmU1JEsM&list=PLTWFMbPFsv...
- i_phish_cats 8y agoYou are evolving the topology, but using regular gradient descent/backprop for any given network, correct?
- jawarner 8y agoNo, in NEAT both the weights and topology are evolved. It is totally gradient-free.
- wholemoley 8y agoYeah, topology and weights. It's highly subject to initial conditions. You almost need another NEAT network to evolve the initial conditions. I believe it's turtles all the way down.