The structure of the BN has been established, and values of probabilities of root nodes (Xi) and each of the conditional probabilities of P(T|Xi) are acquired (T denotes the leaf node).
Are there any methods to learn the conditional probabilities of middle nodes (e.g. P(Bi|Xi), P(Ai|Bi), P(Ci|Xi), P(Bi|Ci))?
How to learn the parameters of middle nodes?
-
- Posts: 2
- Joined: Tue Nov 22, 2016 8:17 am
-
- Site Admin
- Posts: 1417
- Joined: Mon Nov 26, 2007 5:51 pm
Re: How to learn the parameters of middle nodes?
When associating the data with network for EM parameter learning, click on the 'Fixed nodes' button (in the lower left part of the 'Match Network and Data' dialog). Select the nodes for which you want to keep the current parameters; in your case this will be root and leaf nodes.
-
- Posts: 2
- Joined: Tue Nov 22, 2016 8:17 am
Re: How to learn the parameters of middle nodes?
Thank you for your answer, but it's pity that I just learn the parameters from other's thesis, so I can't acquire any date for the parameter learning. So does it mean there is no method to solve it?shooltz wrote:When associating the data with network for EM parameter learning, click on the 'Fixed nodes' button (in the lower left part of the 'Match Network and Data' dialog). Select the nodes for which you want to keep the current parameters; in your case this will be root and leaf nodes.
-
- Site Admin
- Posts: 1417
- Joined: Mon Nov 26, 2007 5:51 pm
Re: How to learn the parameters of middle nodes?
If you don't have any data, you can't perform parameter learning.I can't acquire any date for the parameter learning