Description

EWA's Bayesian Network Engine is used with classification or regression tasks. The engine also incorporated Bayesian Decision Analysis techniques allowing it to make logical decisions, or sets of decisions, based on information encoded in the Bayesian Network. More about the Bayesian Decision Analysis add-on...

EWA's Bayesian Network Engine is used to look for relationship between categorical or binned continuous data. If one thinks of the data in a table format, Bayesian networks assign a graphical node to each column. The presence of arcs connects these nodes denotes the possiblity of a relationship, while the absence denotes no relationship, a much stronger statement. The effect of a relationship between two nodes connected by an arc is encoded in conditional probability table. In EWA's solution, the effect of one node upon another throughout all of the nodes is determined using a process called Join Trees, which is an exact inference method.

The engine's structural learning function then searchs for the best model that represents the given data. This model is represented as a set of arcs that connect the nodes. Between nodes that are not connected (directly or indirectly), there is no discernable relationship, which is very useful to know. Nodes that are directly connected have a relationship that is described by a conditional probability table contained in the node to which is pointed. Nodes that are indirectly related, may or may not be related, depending on the state of evidence in the intervening nodes.

• Implemented in 100% Java, with performance similar to C.
• Unlimited Problem Size (Problem does not have to fit in memory)
• Controlled Multi-Threaded Implementation (Uses only the number of threads specified)
• Full Data I/O Capabilities
  • ODBC and ASCII Data Files
• Handles hard and soft evidence on all nodes.
• Prior and posterior values available on all nodes.
• Structural Learning
  • Entropy- and MDL-based
  • Supports forces and forbidden arcs
• Full Decision Support Package
  • Uncertainty, decision, value, and report nodes supported.
  • Multiple decision and value nodes supported
  • Value-of-deal, Value-of-information, and Value-of-control calculations supported.

• Lumina

• Bayesia

• BayesWare

• Data Digest (Now DecisionQ)

• Inferscape

• Knowledge Industries

• Noetic Systems

• HUGIN

• Netica

Note: EWA Systems wrote the Bayesian Engines behind Inferscape's and Data Digest's solutions.

To Purchase or For More Information, contact our Sales Team.