Search found 1263 matches

by shooltz[BayesFusion]
Mon Oct 26, 2020 2:33 pm
Forum: SMILE
Topic: Error with data set matching for parameter learning
Replies: 1
Views: 13

Re: Error with data set matching for parameter learning

You need to discretize your data before parameter learning if your nodes are discrete. ind_1_5_1 contains values like 46.29 - SMILE can't map this onto a discrete node.
by shooltz[BayesFusion]
Fri Oct 23, 2020 9:24 am
Forum: GeNIe
Topic: Dynamic Bayesian Network (DBN) for time series
Replies: 8
Views: 41

Re: Dynamic Bayesian Network (DBN) for time series

Sorry - the equations are not supported in the DBNs.
by shooltz[BayesFusion]
Mon Sep 28, 2020 12:31 pm
Forum: SMILE
Topic: SMILE 1.6 Released (Unicode fully supported now)
Replies: 0
Views: 532

SMILE 1.6 Released (Unicode fully supported now)

SMILE 1.6 is available now. This version of the library fully supports Unicode in node identifiers, names and other text-based attributes stored in the network.

To download the library, visit https://download.bayesfusion.com

The documentation is available at https://support.bayesfusion.com/docs
by shooltz[BayesFusion]
Wed Sep 02, 2020 10:38 pm
Forum: SMILE
Topic: Accessing cases from JSMILE
Replies: 3
Views: 4255

Re: Accessing cases from JSMILE

Sorry, this feature is not available yet.
by shooltz[BayesFusion]
Wed Sep 02, 2020 10:36 pm
Forum: SMILE
Topic: native library managment and gradle
Replies: 1
Views: 330

Re: native library managment and gradle

We don't have any experience with Gradle; please read the discussion linked below. https://discuss.gradle.org/t/how-to-add-os-native-libraries-to-java-library-path-for-testing/6501 Also note that you can use jsmile.native.library JVM property to specify the full path to libjsmile.so (see https://sup...
by shooltz[BayesFusion]
Mon Aug 31, 2020 9:09 pm
Forum: SMILE
Topic: getNoisyExpandedDefinition() in Dynamic Bayesian Network
Replies: 1
Views: 343

Re: getNoisyExpandedDefinition() in Dynamic Bayesian Network

This feature is not available. There are two possible workarounds. You can either - change noisyMAX node to CPT and use Network.getNodeTemporalDefinition or - keep the noisyMAX type and use Network.unroll to obtain the unrolled network, then locate the unrolled noisyMAX node and call getNoisyExpande...
by shooltz[BayesFusion]
Wed Aug 26, 2020 11:33 am
Forum: GeNIe
Topic: GeNIe 3.0 released
Replies: 0
Views: 652

GeNIe 3.0 released

GeNIe 3.0 is now available at https://download.bayesfusion.com

This build fully supports Unicode (in model labels and filenames). We have also refreshed the user interface.
genie3_scr01_main.png
genie3_scr01_main.png (173.59 KiB) Viewed 652 times
genie3_scr10_unicode_support.png
genie3_scr10_unicode_support.png (73.15 KiB) Viewed 652 times
by shooltz[BayesFusion]
Wed Aug 12, 2020 3:11 pm
Forum: GeNIe
Topic: compare models
Replies: 3
Views: 1486

Re: compare models

The clustering algorithm is the inference algorithm (used to calculate the posterior probabilities). It is applicable once you have the network (both structure and parameters). Its inputs are the network itself and the evidence set in the network. The structure learning algorithms use the data set a...
by shooltz[BayesFusion]
Tue Jul 28, 2020 8:35 am
Forum: SMILE
Topic: How log likelihood value is calculated in EM?
Replies: 6
Views: 1444

Re: How log likelihood value is calculated in EM?

log p(x)=log (product of cliques/product of sepsets)? It's more complicated algorithm. We run relevance decomposition which may create a forest of junction trees (the same approach is used for the main exact inference algorithm). See this post for the explanation of the chain rule algorithm and P(e...
by shooltz[BayesFusion]
Tue Jul 28, 2020 8:14 am
Forum: SMILE
Topic: How log likelihood value is calculated in EM?
Replies: 6
Views: 1444

Re: How log likelihood value is calculated in EM?

The primary algorithm for calculating the probability of evidence is based on the junction tree cliques. If the junction tree cannot be created due to memory constraints, the algorithm switches to chain rule (slower, but potentially less memory-intensive).
by shooltz[BayesFusion]
Fri Jul 24, 2020 9:11 am
Forum: GeNIe
Topic: About Parameter Learning
Replies: 5
Views: 1510

Re: About Parameter Learning

An update: the automatic matching between integer indices (1,2,...) and the corresponding state names (Xyz1, Xyz2, Xyz3...) has been added to GeNIe. We plan to release the next version of the program later this summer.
by shooltz[BayesFusion]
Thu Jul 23, 2020 4:08 pm
Forum: GeNIe
Topic: About Parameter Learning
Replies: 5
Views: 1510

Re: About Parameter Learning

The problem probably is related to the automatic matching between data columns and network nodes. Your data file contains numeric indices. During structure learning these indices are converted into string outcome identifiers. The PC/EM does it internally and there's no issue. Later on, when you run ...
by shooltz[BayesFusion]
Thu Jul 23, 2020 3:17 pm
Forum: GeNIe
Topic: About Parameter Learning
Replies: 5
Views: 1510

Re: About Parameter Learning

Please post your data file and the network (after PC, before EM).
by shooltz[BayesFusion]
Thu Jul 23, 2020 1:39 pm
Forum: SMILE
Topic: pysmile k-fold CV
Replies: 5
Views: 3509

Re: pysmile k-fold CV

Can you share your "validator test.xdsl" file so we can reproduce the problem?
by shooltz[BayesFusion]
Mon Jul 20, 2020 9:26 pm
Forum: SMILE
Topic: pysmile k-fold CV
Replies: 5
Views: 3509

Re: pysmile k-fold CV

SMILE does not support validation (including k-fold) in DBNs. You need to unroll your network and run k-fold on the non-DBN model.