Electronic Statistics Textbook  I  Contact Us  I  Request Quote  I  Search
STATISTICA








New Features in STATISTICA Neural Networks
  • A Quicker, More Simplified Interface

  • New Network types - Self-Organizing Feature Map, Generalized Regression Neural Networks, Linear Models, Cluster Networks and Network Ensembles

  • Training algorithms - Delta Bar Delta, Conjugate Gradient Descent, BFGS, and Quick-Propagation have been added.

  • Overfitting algorithms - early stopping, weight decay regularization, ensemble learning, cross-validation, Monte-Carlo and bootstrap techniques.

  • Improved Intelligent Wizard - automatically selects network types, complexity, and input variables.

  • Ensemble Models and Resampling - the ability to create and store cooperative networks provides a powerful learning tool that can result in a lower generalization error.

  • Comparative results from multiple models - a powerful feature simplifies comparing models in a single spreadsheet or plots the response curves of several models on one graph.

  • Full Integration with STATISTICA - multitasking, simultaneous analyses, workbooks, reports.

  • Visual Basic Scripting - a fully programmable system: users can record interactive analyses as Visual Basic macros for repeated use, access all features from the object model, and integrate STATISTICA with a multitude of other applications such as Excel, C++, Java, etc.

  • Efficient Output Management - STATISTICA features Microsoft Binder-style workbook functionality and easily customizable reports.

  • And many others...

    Back to the STATISTICA Neural Networks page

    Upgrade to STATISTICA Neural Networks 6 Now!

    Request Quote
    StatSoft Home Page


    [StatSoft]
    2300 East 14th Street, Tulsa, OK 74104
    Phone: (918) 749-1119; Fax: (918) 749-2217
    [StatSoft]e-mail: info@statsoft.com

    ©Copyright StatSoft, Inc., 1984-2004.
    StatSoft, StatSoft logo, STATISTICA, SEWSS, SEDAS, Data Miner, SEPATH and GTrees are trademarks of StatSoft, Inc.