Brought to you by Structural Insights®
Latest News

TISEC delivers Practical Pattern Recognition Workshop for NDT in Montreal, Canada, June 28-30, 2010

New look of TISEC's Super ICEPak® web site, November 2009.

Download and try out Super ICEPak® today, contact us for a trial key, November 2009.

Super ICEPak® — the Product

What is Super ICEPak® ?

With Super ICEPak®, Neural Networks, Statistical Pattern Recognition and Fuzzy Logic become practical tools. Furthermore, you don't have to commit to one of these AI tools - they are all available in Super ICEPak® — you can compare each approach with a click of your mouse and see which is best for your application on your own data.

With artificial intelligence in the form of neural networks and statistical pattern classifiers, Super ICEPak® is the most sophisticated program available for multi-parameter interpretation and automated data interpretation. Super ICEPak®'s feature-oriented point-and-click classifier module helps organize, interpret, visualize and analyze data for on-line data interpretation and diagnosis.

Super ICEPak® incorporates a special branch of Artificial Intelligence (AI) called Pattern Recognition (PR) with supervised and unsupervised learning into a package that enables users with little or no prior knowledge of AI and/or PR to use the state of the art technique in the AI field into their own application.

Super ICEPak® uses the classification methods of statistical pattern recognition and neural networks, branches of artificial intelligence (AI) for interpretation and diagnosis. Simulating advanced human functions, it interfaces man and machine to combine their best capabilities. Pattern classification automates the routine interpretation carried out by human analysts. It can be combined with expert systems to produce hybrid AI systems.

Super ICEPak® consists of a standalone Classifier Design Console (CDC) and a set of run-time AI (Dynamic Linked Library) DLLs. The user will first collect data and train the classifier using the Classifier Design Console and then incorporate the finished classifier into their own application with the help of the supplied set of run-time AI DLLs. The run-time AI DLLs consist of two engines &mdash a classification engine and a feature extraction engine. There are five built-in classifier types and six feature sets to choose from the set.

Pattern Recognition and Super ICEPak®

Feature extraction tools are drawn from signal and image processing, or, in the case of other applications, are user defined. The decision processor may be a statistical pattern classifier, or a neural network. The statistical pattern classifiers derive from maximum likelihood theory and neural network depend on operations research optimization theory. The uncertainty is accommodated using fuzzy logic which involves attributing a confidence level to the decision.