Software automates data mining model development.
Press Release Summary:
Available for Windows® 2000 and XP, NeuralSight augments NeuralWorks Predict® platform, which enables users to create neural network models for prediction, classification, and clustering applications. Its GUI handles datasets too large for Microsoft® Excel and allows modelers to specify model performance requirements. Software automatically selects and ranks best neural networks from set of networks built by Predict during unattended model building session.
Original Press Release:
NeuralWare Announces Launch of NeuralSight(R), Automated Data Mining Model Development and Validation Software
PITTSBURGH, Nov. 15 -- NeuralWare, a developer of award- winning desktop neural network data mining software, introduces NeuralSight, the newest member of NeuralWare's family of neural network model development products.
NeuralSight augments and enhances the NeuralWorks Predict(R) platform, which enables researchers and analysts to create robust neural network models for prediction, classification, and clustering applications. NeuralSight extends the power of Predict by providing a Graphical User Interface for handling datasets too large for Microsoft(R) Excel, and by allowing modelers to specify model performance requirements. NeuralSight then automatically selects and ranks the best neural networks from a set of networks built by Predict during an unattended model building session.
Typically, building and evaluating models to analyze large quantities of multivariate data for enterprise applications or analytic/diagnostic devices requires considerable administrative time and effort by the modeler, in addition to the time needed to compute and review performance metrics. NeuralSight automates this process so that hundreds, even thousands, of models can be built and evaluated without modeler intervention, thereby greatly increasing productivity and substantially reducing development costs.
NeuralSight is employed by a leading Japanese bank to streamline its risk management model development process, allowing it to update credit scoring models more frequently while at the same time improving model quality and giving senior loan analysts more time for building customer relationships and identifying new commercial lending opportunities.
NeuralWare has also worked closely in a one-month project with scientists from DuPont Canada to build, test and select models for embedding in a new grain quality assessment system. The application is the DuPont(TM) Acurum(TM) system that DuPont Canada has developed with its strategic partners to enter a new market for grading wheat. Over 5 million data records were processed to yield almost 7,000 models -- each having hundreds of inputs -- before a final ensemble was chosen and validated with independent domain data. Previous product releases took considerably longer, and evaluation was much more cumbersome without the benefit of NeuralSight's automated testing and visualization of model sensitivities to input field variability.
NeuralSight is available now for Microsoft Windows(R) 2000 and Windows XP. Other operating systems will be supported in 2005.
NeuralWare develops sophisticated neural network software solutions for the commercial, scientific, industrial, academic, and government communities. For more information, contact: John Wavle, Vice President for Sales at 412-278-6292, or john.wavle@neuralware.com
Visit NeuralWare on the web at: www.neuralware.com
NeuralWare and NeuralSight are registered trademarks of NeuralWare. DuPont(TM) is a trademark of E.I. Du Pont de Nemours and Company. DuPont Canada is a licensee. Acurum(TM) is a trademark of E.I. Du Pont Canada Company.
