WP 5 - Developments in HUGIN Software Tool

 

OBJECTIVES

  • Development of specific elements of the AMIDST framework in the existing commercial software tool HUGIN to support the analysis of data from the use-cases.
  • Implementation of both new and revised algorithms and methods in the HUGIN tool for scalable data analysis.
  • To support scalable analysis of data and streaming data as defined by the use-cases as well as to scale up existing implementations for data analysis.
     

PLANNED WORK

The work will focus on implementing new algorithms and extending existing algorithms in the HUGIN software tool to support the analysis of data in the use cases. Both algorithms for learning structure as well as inference are to be considered.

Scalability includes parallelization of algorithms for learning Bayesian network structure from data as well as algorithms for constructing structure restricted models. Approximate inference and adaptation of parameters in dynamic Bayesian networks will be considered for analysis of data streams.

Finally, adjustments to the HUGIN software library and Code Wizard will be made to support the automotive use-case.
 


This website uses cookies. Cookies are only used for handling login and identifying registrered users, not for visitor tracking.