BayesFusion Support Forum

No, because in the extreme you will get one state for each value in the data set, which is probably not what you what. You'll have to experiment and somehow evaluate what works best.

Feature selection uses Greedy Thick Thinning (search forum for more info) to first greedily add features and then remove features that maximizes a Bayesian score function.

You can definitely learn these probabilities from data but you'll need time series data.

Is it possible there are no dependencies in the data (i.e., random)?

What kind of noisyMax-specific code are you using for parameter learning? When I did some learning on a noisyMax network (in GeNIe) I got the impression that no special code was used. The reason is that after learning, when I saved the network, the noisyMax parameters in the file where replaced by ...

Hint: The joint distribution is given by P(C,xi,yi,ti,xf,yf,tf)=P(yi)P(xi|yi)P(ti|yi)P(C|ti)P(tf|C)P(xf|C)P(yf|tf). All other probabilities can be derived from that. Does this help?

PC is the only option for continuous data, otherwise you need to discretize.

* What is the new structure represent?
* How can i build BN from my data (with this new structure representation ), when my main purpose is finding dependency between
attributes?
The new structure denotes a direct dependency between variables when an edge exists, and edges are oriented if the ...

1. Your guess is correct.
2. It randomizes all the CPT parameters (but only if the equivalent sample size is zero).

Since they are observed, they don't change when inference is performed. Beliefs of neighboring nodes will be updated.

No, I believe this is still quite different. In case of Bayesian networks only one state can be true at a time but you may not know which one, hence the distributions. In case of a fuzzy Bayesian networks multiple states may be true to certain degrees at the same time, which is conceptually different.

What is a fuzzy Bayesian network?

Hard evidence is observing the state of a node with probability 1. Soft evidence is observing a distribution over the states of a node.

Please search the forum for explanations and references. If you still have questions, please let me know.

For standard greedy search you need a random starting position, this is not necessary for GreedyThickThinning.

Yes, it's a greedy search algorithm that works in two stages.