Cooperative multi-agent reinforcement learning
MattSkala opened this issue · comments
Matouš Skála commented
We want to add support for an authenticated communication between agents, so the bots can share knowledge and converge to the optimal QTable more quickly.
IPv8 will be used for communication and probably Protocol Buffers for message serialization.
Martijn de Vos commented
I would recommend you to use the built-in primitives for message serialization in IPv8, instead of protocol buffers. Please have a look at the existing communities to get an idea how to use them.