Network
Technology
|
Neural Network Technology |
NeuralystTM for Excel |
Neuralyst is a general purpose neural network engine that has been integrated with Microsoft® ExcelTM on WindowsTM or MacintoshTM systems. Neural networks can perform pattern matching or feature recognition, particularly on "noisy" or "fuzzy" data.
Neuralyst provides a friendly user interface and a powerful, flexible neural network that is self-programming. You act as a coach for Neuralyst by providing it with data and letting it know the goals it should learn. Neuralyst will then train itself on the data and goals you have set. While it is training, it will report on how well it is doing.
Once Neuralyst has been trained for a particular application, it can be presented with new data derived from the same or similar sources and it will be able to recognize features consistent with its past learning and advise you on its evaluation or prediction.
Neuralyst also includes an easy to use genetic supervisor which can take a neural network configuration developed by the user and optimize it for the best performance or fastest computation time, using new advanced genetic optimization techniques.
Neural networks have been used for a large number of applications from investment analysis to gambling to image recognition to robotics. The range of applications is limited only by your inventiveness.
Combining the data management, processing, macro and charting capabilities of Excel with a powerful neural network engine places an unprecedented level of processing power at your fingertips. You can develop expert systems, pattern recognizers, outcome predictors, and risk evaluators, all in a functional, friendly user environment.
Excel macros can be used to automate data entry and management. Excel charts can be used to measure neural network characteristics, display neural network goals, or chart success over time. Excel formulas can be used to pre- or post-process data, goals, or evaluations. A fast switch between Excel spreadsheets lets you work with other neural network models, then return to the first. All this within one environment!
Windows or Macintosh system features for multi-tasking and inter-task data transfer allow a multitude of applications and utilities to be integrated with Neuralyst, to deliver neural network technology where you want it in the way you want it.
If you can record your data in spreadsheet or ASCII format, then you can use Neuralyst.
Click on the Neuralyst icon in your Windows or Macintosh system. This will cause both Excel and Neuralyst to load. Neuralyst will be integrated with Excel and its operations will be accessible through two new menus in the Excel menu bar. Read in your data by using Microsoft Excel's File operations. Once your column-organized data is in an Excel spreadsheet then Neuralyst can be applied to the data.
Use the mouse to select the columns to be presented as data. Develop the goals to be matched with the data using Excel's powerful spreadsheet language. With a few more mouse clicks and drags, the entire application becomes defined for Neuralyst. Now, let Neuralyst train on it. The highly optimized neural network engine goes to work and starts learning the relationships and patterns that are inherent between the data and goals that you established. You can track the progress of learning while it occurs as you monitor Neuralyst.
Once Neuralyst has completed learning and has been tested to your satisfaction, it is ready to make evaluations or predictions based on new data!
User Interface - is implemented through an Excel macro package which manages communications between the user and the Neural Network Engine. All Excel features remain accessible and can be used in combination with Neuralyst commands.
Neural Network Engine - is a highly optimized program, written in C, which implements the back-propagation neural network algorithm with selectable activation functions. Computations occur in fixed-point arithmetic or floating point arithmetic at the user's option. On Windows systems it is a Dynamic Link Library (DLL); on Macintosh systems it is a code resource.
Data Types - may be any numeric magnitude or range accepted by Excel. Scaling and normalization is automatically performed by the neural network engine interface. Symbolic data may be represented by text labels.
Input and Output Limits - each Excel column may be designated as an input or output for Neuralyst. The total number of columns used to define an application is limited only by Excel restrictions. Within each column, over 6500 rows may be used for each input or output. The total number of usable columns and rows may be limited by the memory resources available to Excel.
Neural Network Capacity Limits - the neural network may be configured with as many as six layers. The neural network engine will support over 130,000 neural connections. The total number of neurons is limited only by the number of neural connections.
Windows Neural Network Performance - For a 60 MHz Pentium: training mode performance can exceed 440,000 connection updates per second in fixed point; predictive mode performance can exceed 1,360,000 connections per second in fixed point.
Macintosh Neural Network Performance - For a 25 MHz Mac Quadra: training mode performance can exceed 200,000 connection updates per second with fixed point; predictive mode performance can exceed 650,000 connections per second in fixed point. A native PowerMac version is available with even greater performance.
Neuralyst Requirements -For IBM-PC or compatible: Microsoft Windows 3.1 or higher, and Excel 4.0 or higher. For Apple Macintosh: System 7.0 or higher and Excel 4.0 or higher. Neuralyst will run on any system capable of supporting Excel, though faster processors will improve performance and greater memory will increase the size your models can be. Neuralyst itself requires less than 1MB of disk space for its program storage.