Search found 76 matches

by Martijn
Mon May 04, 2015 3:11 am
Forum: SMILE
Topic: Degree of Influence and Sensitivity Analysis
Replies: 18
Views: 23372

Re: Degree of Influence and Sensitivity Analysis

Hi Arsham,

This article should give you an overview of the background of tornado diagrams
http://en.wikipedia.org/wiki/Tornado_diagram

Martijn
by Martijn
Sat Aug 23, 2014 5:41 am
Forum: GeNIe
Topic: sample size meaning
Replies: 2
Views: 3139

Re: sample size meaning

Hi,

The Equivalent Sample Size influences the structure score and is a part of the BDeu calculation, it's sort of represents an inertia of the current parameters when introducing new data.

It is not related to the number of samples in your dataset.

Martijn
by Martijn
Thu Mar 27, 2014 2:47 pm
Forum: GeNIe
Topic: Which parameters influence Bayesian score ?
Replies: 1
Views: 3713

Re: Which parameters influence Bayesian score ?

Hi, These are the parameters that are used for the Bayesian Search algorithm, and I've listed if they influence the score or not (i.e. if the same structure obtained under different parameter settings would result in a different score) Max Parent Count - Limits parents of a node, has no effect on th...
by Martijn
Thu Nov 21, 2013 8:41 pm
Forum: SMILE
Topic: Inference in network
Replies: 3
Views: 3860

Re: Inference in network

Whoops, my bad.
by Martijn
Thu Nov 21, 2013 7:52 pm
Forum: SMILE
Topic: Inference in network
Replies: 3
Views: 3860

Re: Inference in network

Hi Ralph,

It's kept after it's created and is reused for later inferences.

Best,

Martijn
by Martijn
Sat Nov 16, 2013 1:35 pm
Forum: SMILE
Topic: solving undirected arcs
Replies: 3
Views: 3743

Re: solving undirected arcs

Hi Nordic,

We've tracked down a bug causing your problem and have fixed it.
We've updated the GeNIe installer on our downloads page.

After reinstalling everything should be fine.

Best,

Martijn
by Martijn
Thu Nov 14, 2013 11:11 pm
Forum: SMILE
Topic: How to apply filter in DSL_dataset
Replies: 2
Views: 3053

Re: How to apply filter in DSL_dataset

Hi Peter, The filter function is not for filtering variables but for filtering records. If you want to select a subset of variables I recommend making a copy of the dataset and then removing the variables you are not interested in. If I remember correctly, this is how GeNIe does it under the hood fo...
by Martijn
Wed Nov 13, 2013 5:54 pm
Forum: SMILE
Topic: solving undirected arcs
Replies: 3
Views: 3743

Re: solving undirected arcs

Hi Nordic,

It seems your training data is missing, could you repost it.
I'll have a look. I implemented the code responsible for converting patterns into DAGs.
Both GeNIe and SMILE should be using the same code.

Martijn
by Martijn
Tue Nov 05, 2013 11:07 pm
Forum: SMILE
Topic: libstdc++ and libc++ problem in osx 10.9
Replies: 10
Views: 10482

Re: libstdc++ and libc++ problem in osx 10.9

Could you perhaps tell me a bit more about DBN.o and DBNLearner.o?
How were they compiled?
These are from code that you've written?

Martijn
by Martijn
Mon Nov 04, 2013 3:23 pm
Forum: SMILE
Topic: Unsolved External Symbols - Window 7 x64 Visual Studio 2012
Replies: 4
Views: 4856

Re: Unsolved External Symbols - Window 7 x64 Visual Studio 2

I've seen that typo before, it's tutorial code.

What does your main code look like Peter?

Martijn
by Martijn
Wed Oct 23, 2013 6:14 pm
Forum: SMILE
Topic: Expectation-Maximization EM learning
Replies: 1
Views: 3216

Re: Expectation-Maximization EM learning

Hi Bernt, For Bayesian networks we assume Multinomial distributions. Yes, the log value is the loglikelihood of the data, i.e. how well the model fits the data. As long as your data does not contain any missing values the EM algorithm works the same as the same Maximum Likelihood Estimation. If ther...
by Martijn
Fri Oct 18, 2013 3:52 pm
Forum: SMILE
Topic: set and update VS set all evidence and update once
Replies: 2
Views: 3445

Re: set and update VS set all evidence and update once

Hi

In the end, it doesn't make a difference unless you are interested in the intermediate probabilities after setting evidence one by one.
With either method 1 or 2 , if you check the posteriors of the variables after setting all evidence the result is the same.

Best,

Martijn
by Martijn
Wed Oct 09, 2013 6:26 pm
Forum: SMILE
Topic: smile capabilities
Replies: 3
Views: 3731

Re: smile capabilities

Hi Nordic,

No, there currently isn't any.

Best,

Martijn
by Martijn
Wed Sep 25, 2013 3:13 pm
Forum: SMILE
Topic: SMILE not equal results in GENIE
Replies: 2
Views: 3938

Re: SMILE not equal results in GENIE

Hi Bernt, What you want to change is the following line: DSL_sysCoordinates theCoordinates (*theNet.GetNode(mL)->Definition()); Here you are looking at the definition and not the value after updating. It makes sense it's 1.0 because the node is deterministic. Try: DSL_sysCoordinates theCoordinates (...
by Martijn
Sat Jun 08, 2013 3:18 pm
Forum: SMILE
Topic: smile embeded to iphone app
Replies: 4
Views: 5094

Re: smile embeded to iphone app

It seems this might be a more general problem, i.e. how to link to a c++ library from xcode. So it's not really an issue related to smile. You may want to search the web for some instructions. I did a quick google search and I found a few already: http://stackoverflow.com/questions/8950388/link-obje...