Re: BNs from Decision Trees

Stefano Monti 624-8563 (smonti@isp.pitt.edu)
Wed, 23 Sep 1998 10:50:10 -0400 (EDT)

Michael,
you might want to take a look at the work by Mike Jordan on
"probabilistic decision trees", and his work is mainly concerned
with learning these models from data.
Their representation as probabilistic graphical models is briefly
illustrated in

"Hidden Markov decision trees."
ftp://psyche.mit.edu/pub/jordan/hmdt.ps.Z

Their detailed description is in

"A statistical approach to decision tree modeling"
ftp://psyche.mit.edu/pub/jordan/colt94.ps.Z

-- ste

On Wed, 23 Sep 1998, Michael Mayo wrote:

> Here's a question for you all...
>
> If I have a decision tree with probabilities attached to each and
> every outcome arc and leaf nodes representing conclusions, are there
> any algorithms for inducing a bayesian network?
>
> Has anyone done any work on this? Would it be possible to induce a
> bayesian network with a single conclusion node where each value
> corresponds to a leaf node in the DT?
>
> MIke

_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
_/ _/
_/ Stefano Monti, Intelligent Systems Program, University of Pittsburgh _/
_/ _/
_/ 901M CL, Pittsburgh, PA - 15260 _/
_/ http://www.isp.pitt.edu/~smonti phone: (412) 624-8563 _/
_/ smonti@isp.pitt.edu fax: (412) 624-6089 _/
_/ _/
_/ Non c'è nulla di meno filosofico che volere savia e filosofica tutta _/
_/ la vita." -- Giacomo Leopardi _/
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/