Hello,

I was wondering how it is possible to perform layer-by-layer learning bottom-up using OpenBayes. For example, there is bottom layer of "sensory" nodes which are fed the data, and want to develop a tree structure where the higher level nodes capture coincident occurrences at lower level nodes.

How would I go about it?

Regards

could someone please provide me some help with this? appreciate your help

Hi aaron.

I'm not sure I understand your question. Could you please give us some more details?

kosta

Kosta,

I am trying to emerge a tree structured Bayesian model from data. The sensory data is from the lowest layer nodes - say temperature, pressure and humidity.

Say Temp_Press is the node the captures co-incidences between temperature and pressure nodes, while press_humid is the node that captures coincidences between pressure and humidity.

The top-most node in the tree (Temp_Press_Humid) captures coincidences between nodes "Temp_Press" and "Press_Humid".


Now I have measurements for temperature, pressure and humidity and I wish to populate the CPT for this model. How would I go about it?

I hope that made some sense..

Regards
aaron

here is a crude drawing:

TPH
/ \
TP PH
/ \ / \
T P H