hi,
I am interested in using SMILE for assisting in making real-time decisions.
That is, can the influence diagram be updated at a "real-time" rate so that the decision making can be done in real-time? OR
is it necessary to augment SMILE with other components such as Hidden Markov Model, etc.
Thank you so much for your help!!!
Cheers,
Ed Sykes
Real-time Decision Making
-
- Site Admin
- Posts: 1417
- Joined: Mon Nov 26, 2007 5:51 pm
Re: Real-time Decision Making
That's rather general question. The performance will depend on the structure of the model and actual decision set - it's hard to estimate if given network will fit your definition of real time.
-
- Site Admin
- Posts: 1417
- Joined: Mon Nov 26, 2007 5:51 pm
Can't really tell without seeing the network (or, in fact, running the inference in tight loopsykes wrote:Suppose I had a BN of 30 nodes, say 20 of which are constantly being updated....
The number of nodes is not the deciding factor in itself when it comes to performance. By '20 nodes being updated' above you mean calling SetDecision on 20 decision nodes, then doing inference with UpdateBeliefs?
"By '20 nodes being updated' above you mean calling SetDecision on 20
decision nodes, then doing inference with UpdateBeliefs?"
Right.
I am hoping to draw input from Second Life client as the user goes through a set of activities, feed the information to 20 nodes, then doing inference on the newly updated values in the nodes....
Link to Second Life:
http://secondlife.com/
Second Life source is C/C++ :
https://wiki.secondlife.com/wiki/Source_downloads
Thank you very much for your help!!!
Cheers,
Ed
decision nodes, then doing inference with UpdateBeliefs?"
Right.
I am hoping to draw input from Second Life client as the user goes through a set of activities, feed the information to 20 nodes, then doing inference on the newly updated values in the nodes....
Link to Second Life:
http://secondlife.com/
Second Life source is C/C++ :
https://wiki.secondlife.com/wiki/Source_downloads
Thank you very much for your help!!!
Cheers,
Ed