||This article contains content that is written like an advertisement. (February 2016) (Learn how and when to remove this template message)|
Neural Lab is a free neural network simulator that designs and trains artificial neural networks for use in engineering, business, computer science and technology. It integrates with Microsoft Visual Studio using C (Win32 - Wintempla) to incorporate artificial neural networks into custom applications, research simulations or end user interfaces.
It provides a visual environment to design and test artificial neural networks.
The latest Neural Lab version is 4.1. The two major versions are version 3.1 and 4.0.
Version 3.1 is navigated using a standard computer mouse. Version 3.1 is considered easier to use, however, it is difficult to perform complex tasks programmatically. Version 3.1 is therefore primarily useful for people without a programming background.
The version 3.1 tutorial provided very little theoretical background on artificial neural networks. Despite the number of examples, most of the examples focus only on multi-layer networks with supervised training.
In version 4.0, the authors incorporate background information on artificial neural networks.
Version 4.0 incorporates Kohonen networks that can be trained without supervision and probabilistic neural networks.
- The tools allow reviewing and analyzing the structure of the training set.
- The activation of the neurons for each case in the data set are visible. The tutorial provides examples in prediction, data mapping, data classification and autoassociative memory problems.
- Once a network has been trained, it is possible to save it to a file. The file can be opened using Microsoft Visual Studio to create a standalone application that can employ the network.
Specific examples of neural networks include:
- Auto Association
- Network Simulation
Neural Lab is developed using Wintempla (a plug in that works with Microsoft Visual Studio). Wintempla encapsulates Win32 and simplifies the development of Microsoft Windows applications using C++ and native Win32 APIs.
Wintempla is a tool that integrates with Microsoft Visual Studio. Wintempla encapsulates Win32 to simplify the creation of Web and Desktop applications using C++ and object-oriented programming. The programmer has the option to use the native Win32 APIs or the Wintempla classes.
- A tutorial with key concepts in programming
- Videos to illustrates how common control instructions (such as: if, else, for, while, etc.) work
- Many examples and problems that can be used in: programming classes, SQL, PLSQL, Graphics
- Support to create SQL database applications
- SQL Import to create (in seconds) desktop or web applications from a SQL file
- Simulated annealing optimization
- Genetic algorithm optimization
- Asynchronous module for Digital to Analog converters (DAC)
- Asynchronous module for Analog to Digital converters (ADC)
- Asynchronous module for serial ports
- Multithread applications
- Document printing
- Microsoft Windows services
- GUI deployment
- Digital Signal Processing (remez, FFT and Filtering)
- Common Object Model (COM)
- A Lexical Analyzer, a compiler and virtual Machine
- Artificial Neural Networks
- Matrix operations
- Data Visualization: Pie Chart, XY Chart, Polar Chart, Histogram, 3D Visualization, Simulation View
- Native support for string manipulation using the STL
- Native support for Math operations using the STL
- Native support for data file storage
- GDI Game application
- Support for DirectX applications
- Support for Open GL applications
- Support to create PDF files programmatically
- .lab Neural Lab code (a UNICODE text file)
- .lay A multi-layer neural network file
- .lax A complex-domain multi-layer neural network file
- .koh A Kohonen neural network file
- .prb A probabilistic neural network file
- .csv A comma separated data file
- Masters, Timothy (25 July 1994). Signal and Image Processing with Neural Networks: A C++ Sourcebook. John Wiley & Sons. ISBN 978-0-471-04963-0.
- Masters, Timothy (17 April 1995). Advanced algorithms for neural networks: a C++ sourcebook. Wiley. ISBN 978-0-471-10588-6.