5 ms·
Other networks learn - this network evolved. By breeding the winners of many self-play tournaments a perfect player was born. Training and learning show examp
by deepnet 11y ago
Other networks learn - this network evolved.
By breeding the winners of many self-play tournaments a perfect player was born.
Training and learning show examples and mark the student against a correct answer - the student network tries to minimise their error.
Genetic Algorithms are a type of Reinforcement Learning, i.e. learning by trial and error.
The agent can direct it's own learning and explore.
Useful for training robots and other embodied agents such as oppenents.
- discardorama 11y agoI am aware of all of those terms. But I'd prefer to hear details about how this particular network was evolved, as you put it.
- deepnet 11y agoPretty comprehensive details here. http://blog.otoro.net/2015/03/28/neural-slime-volleyball/ http://blog.otoro.net/2015/03/28/neural-slime-volleyball/
- discardorama 11y agoThanks! That's what I was looking for. Very interesting.