Darwin  1.10(beta)
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Groups Pages
Graphical User Interface for Machine Learning Data Flows

This project implements an interface for graphical development of machine learning data flow nextorks. A data flow network is a set of data processing nodes connected in an acyclic graph. Data (in the form of records) moves from source nodes, through processing nodes, to sink nodes. Visualization and debugging nodes provide tools for algorithm introspection. An example data flow network (within the GUI) is shown below.


The wxWidgets GUI toolkit needs to be installed for this project. See Linux Installation Instructions or Windows Installation Instructions for installation instructions.

This project is experimental. Not all features are fully implemented or supported.

An example data flow network for classification is provided in projects/gui/examples/exampleDataFlowNetwork.xml. You can open the network from the command-line with

bin/darwin projects/gui/examples/exampleDataFlowNetwork.xml

There are four basic operations that you can perform on a network (or subnetwork by selecting a subset of the nodes). The operations are:

The Initialize (F5) option from the Network menu will iterate throught nodes in the network (in topological order), first estimating the parameters for the block and then evaluating forwards.

Nodes are divided into six functional groups:

See Also
Applications and Project Descriptions