I have a project that calls for some Bayesian inference. I checked out
the CMU AI repository, but all of such libraries are deprecated.
This, of course, is a great opportunity to build a Bayesian inference
library myself. However, it will still be very helpful to reuse some
existing components.
Can someone point me to the right direction? Thanks a lot.
On Apr 23, 7:03 am, JT <···········@gmail.com> wrote:
> I have a project that calls for some Bayesian inference. I checked out
> the CMU AI repository, but all of such libraries are deprecated.
>
> This, of course, is a great opportunity to build a Bayesian inference
> library myself. However, it will still be very helpful to reuse some
> existing components.
>
> Can someone point me to the right direction? Thanks a lot.
I saw CL code on github recently.
(for the others -- bayesian networks are graphical-models, i.e. data
analysis using network representation of variables -- not bayesian
inference per-say, but a means of establishing directional correlation
(a weird definition for "causal").
> I saw CL code on github recently.
>
> (for the others -- bayesian networks are graphical-models, i.e. data
> analysis using network representation of variables -- not bayesian
> inference per-say, but a means of establishing directional correlation
> (a weird definition for "causal").
Thanks for the tip. You also made me realize that I made a mistake
with my terminologies.
To clarify, I want a "Bayesian Network library", not "Bayesian
inference code", which can be easily written.
JT <···········@gmail.com> writes:
> I have a project that calls for some Bayesian inference. I checked out
> the CMU AI repository, but all of such libraries are deprecated.
How can an algorithm become deprecated?
> This, of course, is a great opportunity to build a Bayesian inference
> library myself. However, it will still be very helpful to reuse some
> existing components.
>
> Can someone point me to the right direction? Thanks a lot.
If you have the Baysian code you need in the CMU AI Rep, go ahead, use it!
--
__Pascal Bourguignon__
On Apr 22, 10:53 pm, ····@informatimago.com (Pascal J. Bourguignon)
wrote:
> JT <···········@gmail.com> writes:
> > I have a project that calls for some Bayesian inference. I checked out
> > the CMU AI repository, but all of such libraries are deprecated.
>
> How can an algorithm become deprecated?
>
> > This, of course, is a great opportunity to build a Bayesian inference
> > library myself. However, it will still be very helpful to reuse some
> > existing components.
>
> > Can someone point me to the right direction? Thanks a lot.
>
> If you have the Baysian code you need in the CMU AI Rep, go ahead, use it!
What I mean was, the codes listed by CMU AI rep does not exist any
more. (At least I can't find them.)
JT <···········@gmail.com> writes:
> What I mean was, the codes listed by CMU AI rep does not exist any
> more. (At least I can't find them.)
I was able to find the code without too much trouble. The only slighly
obscure part is when you have to click on a link to get a directory view
that shows you the actual files you can download.
http://www.cs.cmu.edu/afs/cs.cmu.edu/project/ai-repository/ai/0.html
Where were you looking?
--
Thomas A. Russ, USC/Information Sciences Institute
Ok, that's stupid of me. I missed this one somehow. But it's the only
lisp Bayesian network code I can find.
http://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/areas/reasonng/probabl/belief/0.html
I will look into it and see what I can use.
On Apr 23, 5:52 pm, ····@sevak.isi.edu (Thomas A. Russ) wrote:
> JT <···········@gmail.com> writes:
> > What I mean was, the codes listed by CMU AI rep does not exist any
> > more. (At least I can't find them.)
>
> I was able to find the code without too much trouble. The only slighly
> obscure part is when you have to click on a link to get a directory view
> that shows you the actual files you can download.
>
> http://www.cs.cmu.edu/afs/cs.cmu.edu/project/ai-repository/ai/0.html
>
> Where were you looking?
>
> --
> Thomas A. Russ, USC/Information Sciences Institute
IME, most of the actual code links pointed to are dead. Try
downloading some and you'll see what he means.
On Wed, 22 Apr 2009 22:03:36 -0700, JT wrote:
> I have a project that calls for some Bayesian inference. I checked out
> the CMU AI repository, but all of such libraries are deprecated.
>
> This, of course, is a great opportunity to build a Bayesian inference
> library myself. However, it will still be very helpful to reuse some
> existing components.
>
> Can someone point me to the right direction? Thanks a lot.
What exactly do you need? The posterior function usually needs to be
written by you, unless you are using some well-known family of models,
sometimes the latter are prepackaged, but it is not a big deal to
write them anyway. As for sampling, writing a Metropolis-Hastings or
a Gibbs sampler should be pretty straightforward too, and the latter
should be adapted to your model, so I don't see how much of it could
end up in a library.
Tamas