Outstar learning law (Grossberg, 1976) governs the dynamics of feedback connection weights in a standard competitive neural network in an unsupervised manner. This learning models how a neuron can learn a top-down template corresponding to, i.e., expect, a particular input pattern.
Below are links to source article, tutorial, and zipped file that contains a MATLAB-based graphical user interface with additional access to the outstar learning law equation, description, and source code.
The microcircuit for the outstar learning law shows how the dynamics of feedback weights from nodes in a coding field to nodes in an input field are governed within a standard competitive neural network in an unsupervised manner (Grossberg, 1976). This learning models how a neuron in the brain can learn a top-down template corresponding to a particular input pattern. An example simulation allows the users to see how the outstar learning law (Grossberg, 1976) changes weights of connections from the winning node at a coding field that diverge onto an input field. With outstar learning, these efferent weights eventually learn to expect the input activation pattern. This law incorporates Hebbian learning and pre-synaptically gated decay. Typically, learning occurs only for weights that diverge from active nodes in the coding field. However, learning can be further confined to weights projecting away from the most active node in the coding field assuming winner-taking-all coding in the network. This is called competitive learning.
[ http://techlab.bu.edu/MODE/outstar_tutorial.ppt ] The tutorial is a self-contained power point presentation that introduces the outstar learning law.
To use the software for the outstar learning law, download the package (Outstar_GUI_070109.zip) from the Download(s) below and unzip the contents into a local folder. Open MATLAB and change the current directory to the folder. At the command prompt, type outstargui to begin using the software via a GUI.
Any operating system that can support MATLAB
Praveen K. Pilly