Privateer™

Product Summary

A fast, versatile neural network trainer block for the Simulink package from Mathworks, capable of modeling very complex systems built from Simulink blocks and replacing them with a smaller memory footprint simulations.  Network parameters are outlined in an XML script.  Privateer requires Simulink and an IEI security dongle to function.

Product Features

• XML-based setup for training parameters and file i/o
• Allows for input and output definition using Simulink’s drag-and-drop interface.
• Allows for any number or size of systems to be modeled.
• Networks can be used in series to create powerful neural cascades within Simulink.

Details

Privateer is a Neural Network object for the Simulink package from Mathworks.  This product brings IEI’s STANNO technology into the Simulink world, allowing users to implement neural network representations of extremely complex systems with a fraction of the memory footprint and 'hassle factor'. 

To use Privateer, you must first define your neural network parameters.  Such network parameters are outlined using a simple XML scripting language (similar to that employed by IEI's PatternMaster) that allows for maximum flexibility and code reuse.  Training is accomplished by simply connecting the inputs and outputs of the target system to the Privateer block and running the simulation. Once Privateer has absorbed the behavior of the target system, you may return to your XML script and modify it to turn off the learning mode.  Once the block is reloaded, you can use the trained network block to completely replace the learned system.