Hi,
I am just wondering if anyone is somewhat familiar with the above topics and might be able to provide some good source of information.
So far, I've only really learnt about the "what" and the "why".. But not much about the "how".
Hi,
I am just wondering if anyone is somewhat familiar with the above topics and might be able to provide some good source of information.
So far, I've only really learnt about the "what" and the "why".. But not much about the "how".
I completed a B.Sc. in Artificial Intelligence and a Masters in Cognitive Science, so I guess I should know something or other, but it was a few years back now.
The term (agents) is bandied about quite a lot and often misused. I think a suitable definition is a piece of software that, in some manner, behaves autonomously within a system. I.e. you could not predict what the agent will do next unless you had a complete knowledge of both the agent and the system it operates in.
How? There are many ways of implementing them. The basic model is how I described above. You have your agent and a world that it lives in. The agent receives information about the world and, based on this information and its knowledge, performs an action.
For example, for my B.Sc thesis I wrote an agent based car racing simulator. 4 cars (each one an agent) were placed on a track (the world) and had to race each other, trying to complete the circuit first. Each car had a neural network which "learnt" how to drive based on reinforcement learning (quite simply encouraging good "behaviour" and discouraging "bad" behaviour).
That's pretty much all there is to it. Obviously as the world and the agents increase in complexity things get more complex, but the basic "how" remains the same.
"All our beliefs are being challenged now, and rightfully so, they're stupid." - Bill Hicks
There are currently 1 users browsing this thread. (0 members and 1 guests)