New Release - Hugin 6.8

• A new HUGIN .NET API is introduced with this release
• 64-bit versions of the HUGIN Graphical User Interface and most APIs
• Support for Unicode in the HUGIN software (Graphical User Interface, Java and .NET APIs)
• A Data Analysis Wizard in the HUGIN Graphical User Interface
• Value of information analysis on decisions in influence diagrams in the HUGIN Graphical User Interface
• Hypothesis driven conflict analysis in the HUGIN Graphical User Interface

• The Value of Information (VOI) Dialog has been extended to include VOI analysis on decisions in influence diagrams. Value of information analysis on a decision variable may help the decision maker to determine if an additional observation on a discrete chance node will improve the maximum expected utility of the decision under consideration.

• The Conflict Analysis Dialog has been extend with support for hypothesis driven conflict analysis on all states or a single state. 

• The Evidence Sensitivity Analysis dialog now has a progress bar for calculations that require a large number of propagations. This includes the option to terminate the calculation before all propagations are performed.
• It is possible to discriminate between two hypotheses corresponding to states of the same hypothesis variable.

• A Convolution and XOI loss model dialog for computing total impact based on the exposure, occurrence, and impact model has been included. 

• The HUGIN Graphical User Interface now supports the use of class collections. It is possible to load and save a class collection. A class collection contains a set of class definitions. Class collections allow the user to store multiple class definitions in a single file.
• A new tool, called "View Component Tree", gives a graphical overview of all classes instantiated in a single network class and its subclasses, etc.
• The HUGIN Graphical User Interface reports the resulting number of states of a chance or decision node when adding states. Also, the number of states of a selected decision node is reported.
• Possibility to view probabilities in node tables with a forced number of decimals instead of scientific notation.
• A new "Case Manager Tool" makes it easy to manage and switch between multiple evidence scenarios in a network. The tool supports loading and saving of multiple cases as a HUGIN data file.
• An option to relax the type checking on bounding links (i.e., a link from a node to an input node) has been included. This allows the user to choose to have the type checking performed at compile time. Default behaviour is to perform type checking when a link is added to the model.
• In the Learning Wizard a minimal solution to an ambiguous region consists of a number of uncertain edges. If there are multiple minimal solutions to an ambiguous region, the user has to select one of the possible solutions. To guide this selection the weight and p-value of an uncertain edge is displayed in the status bar when the uncertain edge is selected.
• Monitor windows for continuous chance nodes now support retract evidence.
• GUI response when editing large CPT's have been speeded up.
• A simple tool - The States Generator - for generating and coping states of discrete nodes has been developed. This tool is particularly useful for specifying states of numbered and interval nodes.
• The continuous-discrete node value transfer function (CDVT function) implements support for linking a continuous node to a interval chance node. The table for the interval chance node will be generated by the tool from an expression equal to the posterior marginal distribution of the continuous node. This implement a simple method for discrete children of continuous nodes using discretization with support for forward reasoning along the CDVT function link.
• The "Enter Likelihood Evidence" dialog shows labels or state values depending on the selection of the user in the table of the corresponding node.
• When a numbered discrete chance node is selected in Run-Mode, the status bar will show the average of the node. All state values of the node should be finite.
• The compilation process reports the approximate and compression rate to usage log.
• Unicode support is offered with respect to node and state labels as well as filenames. Node names do not include support for unicode.
• This release include 64-bit versions of the HUGIN Graphical User Interface. The 64-bit version is offered as a separate package with 64-bit versions of the HUGIN C/C++/.NET/Java APIs.

Finally, efforts have been put into improving the stability of the HUGIN Graphical User Interface.

• A new .NET API is now available for the Windows platforms. This API targets the .NET 2.0 framework.
• API libraries for the Windows platforms are now provided for Visual Studio 2005 (in addition to Visual Studio 6.0 and Visual Studio .NET 2003).
• The APIs (except the Visual Basic API) are now available as 64-bit versions on all platforms except Mac OS X.  On the Windows platforms, the 64-bit libraries are only provided for Visual Studio 2005.
• A new naming scheme has been introduced for the API libraries on the Windows platforms: The libraries are now uniquely named, making it possible to have all DLLs in the search path simultaneously.
• d-separation analysis is now used to improve the performance of inference. This is particularly useful for incremental propagation of evidence in large networks.
• The performance of the total-weight triangulation method has been greatly improved.
• The triangulation functions now construct junction trees, but do not allocate storage for the data arrays of the clique and separator tables. This permits the application to see the junction trees before attempting the final (and most expensive) part of the compilation process.
• It is now possible to query the size of a junction tree (even before storage is allocated for the junction tree tables).  
• A function for testing whether a domain is triangulated is now provided.
• The HUGIN KB file format has changed (in order to handle HKB files produced by 64-bit versions of the API, among other things).

• If a compiled (but not compressed) domain is saved as an HKB file, it will only be triangulated when loaded. A compilation is required before inference can be performed. Compressed domains are still loaded as compressed (which implies compiled), but a propagation is required before beliefs can be retrieved.
• Functions for converting between table indexes and state configurations are now provided.
• A function to retrieve the CG size of a table is provided.