Search found 430 matches

by marek [BayesFusion]
Mon May 15, 2017 8:12 pm
Forum: SMILE
Topic: MCMC
Replies: 4
Views: 5045

Re: MCMC

I'm afraid not. Neither among the belief updating nor the learning algorithms. I'm not sure about the performance of learning based on MCMC but in case of belief updating, importance sampling is much better in practice. We suggest that you use the EPIS-BN algorithm.
I hope this helps,

Marek
by marek [BayesFusion]
Mon May 08, 2017 7:44 pm
Forum: GeNIe
Topic: output error message
Replies: 1
Views: 2678

Re: output error message

This error may be a sign of loss of precision. If so, it happens in very large or complex models. We will be able to verify this if you share your model with us. If you don't want to post it on the Forum, please send it to us privately.
Cheers,

Marek
by marek [BayesFusion]
Fri Apr 28, 2017 4:28 pm
Forum: GeNIe
Topic: My problems in using GeNIe as diagnostic test tool
Replies: 1
Views: 2648

Re: My problems in using GeNIe as diagnostic test tool

Rasha,

I will be glad to check what is happening in your model. Will you be willing to share it with me at this stage? You can send it to me in private if you would rather not post it with your message.
Cheers,

Marek
by marek [BayesFusion]
Thu Apr 27, 2017 7:18 pm
Forum: GeNIe
Topic: Elicitate Function in Equation Nodes
Replies: 1
Views: 2686

Re: Elicitate Function in Equation Nodes

We are in the middle of developing a complete hybrid Bayesian networks capability. The newest release, to appear in mid-May, will allow you to create models consisting of a combination of discrete and continuous nodes. May I ask you to have patience with us and wait a little with this bug? It will b...
by marek [BayesFusion]
Tue Apr 18, 2017 3:58 pm
Forum: SMILE
Topic: Simulation real time
Replies: 2
Views: 3641

Re: Simulation real time

Hi, GeNIe's limitation is just for your own sake :-) -- 1000 slices is possibly a computational challenge but it could work, depending on the complexity of your network. Coming back to your main question whether you should model only slices during which something happens or all slices, it depends on...
by marek [BayesFusion]
Fri Apr 14, 2017 7:47 am
Forum: SMILE
Topic: Weight and interval parameters
Replies: 5
Views: 5435

Re: Weight and interval parameters

OK, I'm starting to understand what you want to do. It seems that you want to build a qualitative probabilistic network (like a QPN) with signs of influences being propagated rather than numbers. GeNIe and SMILE are quantitative, i.e., they require a full numerical specification of the joint probabi...
by marek [BayesFusion]
Wed Apr 12, 2017 9:32 pm
Forum: SMILE
Topic: Weight and interval parameters
Replies: 5
Views: 5435

Re: Weight and interval parameters

Could you please clarify your question? What do you mean by images? What do you mean by weights? There are several parameters that can be called weights, e.g., parameters in linear models, weight in utility functions, etc. Do you mean specification of interval rather than point probabilities? Cheers...
by marek [BayesFusion]
Fri Apr 07, 2017 9:41 am
Forum: GeNIe
Topic: [Basic] - How to use beta density functions in equation nodes correctly
Replies: 5
Views: 5880

Re: [Basic] - How to use beta density functions in equation nodes correctly

You are right about the X1 = 1 + Bernoulli (1-F11): I have added 1 in order to shift the result from the interval [0;1] to the interval [1;2]. The 1-F11 and 1-F12 are because Beta distributions in the model are for the probability of heads and you have defined heads as 1 and tails as 2. This is flip...
by marek [BayesFusion]
Sat Apr 01, 2017 9:28 pm
Forum: GeNIe
Topic: [Basic] - How to use beta density functions in equation nodes correctly
Replies: 5
Views: 5880

Re: [Basic] - How to use beta density functions in equation nodes correctly

Thank you for the additional explanation and the reference to Neapolitan's book. I believe I understand your problem better now. What looks like a BN in Neapolitan's book is not really a BN. He wants to show how the parameters of X1 and X2 are calculated based on the Beta distributions (gray nodes)....
by marek [BayesFusion]
Fri Mar 31, 2017 7:40 pm
Forum: GeNIe
Topic: Learned posterior probabilities
Replies: 1
Views: 2509

Re: Learned posterior probabilities

I will be glad to help here. Please create the probability tables based on your knowledge of the interactions between the parents and the children. It is OK to make these approximate. If you know the nature of the relationships between the parents and the children in terms of equations, there is a w...
by marek [BayesFusion]
Fri Mar 31, 2017 7:30 pm
Forum: GeNIe
Topic: [Basic] - How to use beta density functions in equation nodes correctly
Replies: 5
Views: 5880

Re: [Basic] - How to use beta density functions in equation nodes correctly

I will be glad to try to help you. However, I would like to ask you first what the variables X1 and X2 represent. Clearly, variable F11 represents the probability of obtaining a head in a single coin toss. Variables F21 and F22 represent the probability of obtaining a head in a single toss of a coin...
by marek [BayesFusion]
Sat Mar 25, 2017 1:43 pm
Forum: GeNIe
Topic: probability adapting
Replies: 1
Views: 2589

Re: probability adapting

As far a static network goes, I think the way to do it is to create a node Q with several outcomes, e.g., Q0, Q1, Q2_5, Q6_10, each of which denotes the number of times Q has occurred. Then, you enter evidence for the appropriate state. If Q occurred 4 times, for example, the observed state of Q is ...
by marek [BayesFusion]
Sun Mar 12, 2017 8:03 pm
Forum: SMILE
Topic: Issues with node states flip in parameter learning output
Replies: 14
Views: 11764

Re: Issues with node states flip in parameter learning output

[Edited Monday, 13 March 2017] The behavior that you described in your original post is correct (different distributions for the latent variables when starting from different random priors). Latent variables have a lot of freedom in their probability distributions. As long as the joint p.d. over the...
by marek [BayesFusion]
Mon Jan 23, 2017 11:09 pm
Forum: GeNIe
Topic: How to create higher order markov chain model?
Replies: 4
Views: 4757

Re: How to create higher order markov chain model?

Just wanted to add that you may want to add higher level influences than 1. If you just have influence of degree 1, you will have by definition a first order Markov model. Adding higher order influences will allow you to specify a higher order model. I second Tomek's advice that you study the DBN ch...
by marek [BayesFusion]
Mon Jan 23, 2017 11:04 pm
Forum: GeNIe
Topic: Problem with time count.
Replies: 2
Views: 3040

Re: Problem with time count.

I just wanted to add that you may want to see the posteriors over a smaller number of slices. In my experience, the posteriors over different dime slices very quickly converge and don't change much. Perhaps 500 slices is a little too much? 50 slices may show you what you want. Please try "Tempo...