search

bayesnet / bnt

Bayes Net Toolbox for Matlab
460 stars 247 forks source link

learncg.pdf. #81

Open lhw-kl opened 4 years ago

lhw-kl commented 4 years ago

I cant't find the file"www.ai.mit.edu/~murphyk/Papers/learncg.pdf.", if anyone who has this file to share,I would be very thankfull.

murphyk commented 4 years ago

The details are in my thesis https://www.cs.ubc.ca/~murphyk/Thesis/thesis.html

On Wed, Jul 22, 2020 at 6:53 PM lhw-kl notifications@github.com wrote:

I cant't find the file"www.ai.mit.edu/~murphyk/Papers/learncg.pdf.", if anyone who has this file to share,I would be very thankfull.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/bayesnet/bnt/issues/81, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDK6EC5NSZW6DV54WVBCYTR46JY7ANCNFSM4PFIR3RA .

lhw-kl commented 4 years ago

The details are in my thesis https://www.cs.ubc.ca/~murphyk/Thesis/thesis.html … On Wed, Jul 22, 2020 at 6:53 PM lhw-kl @.***> wrote: I cant't find the file"www.ai.mit.edu/~murphyk/Papers/learncg.pdf.", if anyone who has this file to share,I would be very thankfull. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#81>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDK6EC5NSZW6DV54WVBCYTR46JY7ANCNFSM4PFIR3RA .

It's a great honor to recerive your reply so soon. I've downloaded your thesis and will read it carefully.

lhw-kl commented 3 years ago

@murphyk I’ve read about the Conditional Linear Gaussian in your thesis and the relevant code in the BNT, and I have some questions: 1) Is the discrete-valued variable as a part of the training set or not when parameter learning? 2) After using the function ‘[bnet2,LL,engine2]=learn_params_em(engine,cases,max_iter,thresh)’ to estimate the parameter, when query with the function ‘s = struct(bnet.CPD{i})’ and ‘s2 = struct(bnet2.CPD{i})’, why got the same ‘weights’? Such as: ![Uploading image.png…]() Looking forward to your guidance.