smile capabilities

[nordic]

Hello,

Is there any build in Funktion to build an bayesian multinet with smile or Genie (just like Naive Bayes, TAN, Augmented naive Bayes etc...)? Or do discriminative parameter learning?

Best regards,
nordic

[Martijn]

Hi Nordic,

No, there currently isn't any.

Best,

Martijn

[marek [BayesFusion]]

Martijn is right -- we don't have any built-in facilities. However, you should be able to implement them on top of SMILE, which especially for simple structures like Naive Bayes should not be hard. If you tell us more of what you want perhaps we can help.
Cheers,

Marek

[nordic]

thanks, for the quick requests.
I'm just starting to use the smile framework to build an classifier for objects in images. In the Literatur there are some papers which show that learning an bayesian network discriminatively for classification tasks could improve it's correctness. So i wondered if there might be an build-in facility to do this... Obviously there isn't, so im using the "normal" generative approach for now...maybe it's already sufficient...I just wanted stay sure I'm not overlooking some feature and may regret it later