3

I am talking about relationships between AIs (e.g. 2 of them forming a couple, 3+ in family like relationship).

What knowledge could come out of such experimentation?

Matthew Gray
  • 4,252
  • 17
  • 27
K_Rol
  • 131
  • 3
  • 2
    Does [swarm robotics](https://en.wikipedia.org/wiki/Swarm_robotics) fit your question? – Basile Starynkevitch Nov 02 '16 at 18:58
  • If there is a single/common goal, how do define the boundaries between the AIs if they are integrated in one system? If there are separate goals for each of the AIs then what is the interaction for? – Iliyan Bobev Nov 02 '16 at 19:27
  • 2
    https://en.wikipedia.org/wiki/Generative_adversarial_networks – BlindKungFuMaster Nov 03 '16 at 09:04
  • @Basile Starynkevitch: Yes definitely! I would like to get an input on that as well. I actually did not think of it at first. – K_Rol Nov 03 '16 at 14:38
  • For physical example, see: [How can thousand-robot swarm coordinate their moves without bumping into each other?](http://ai.stackexchange.com/q/1426/8) – kenorb Nov 04 '16 at 11:19
  • Similar: [Which deep networks are scalable in terms of “talking” to each other?](http://ai.stackexchange.com/q/160/8) – kenorb Nov 04 '16 at 11:23

1 Answers1

4

One example might be self-play in games. Since neural networks and deep learning depend on massive amounts of data, one way to generate data is to have two virtual machines play each other and record the experience. An example discussion can be found at http://www.cs.cornell.edu/boom/2001sp/Tsinteris/gammon.htm which uses reinforcement learning. I believe AlphaGo also uses this technique of self-play, and uses two independent neural networks, one reducing the search space and the other deciding on the best move in the remaining space, that in a sense cooperate to decide on the next move.

Colin Beckingham
  • 367
  • 2
  • 8