Neural Lab

Neural Lab is a free neural network simulator to design and train artificial neural networks for engineering, business, computer science and technology. It integrates with Microsoft Visual Studio with C (Win32 - Wintempla) to incorporate Artificial Neural Networks in custom applications, research simulations or end user interfaces.

Version 3.x and Version 4.x Functionality Differences

The current version is 4.1. There are two major versions: version 3.1 and 4.0. Version 3.1 provides most of its functionality by using the mouse. Version 3.1 has the advantage that it is easy to use. however, it has the disadvantage that the user cannot perform complex task programmatically. Because the training of an artificial neural network might take several hours, version 3.1 is only useful for people without a programming background. In version 4.0, it is possible to perform artificial neural network operations by writing code. The code is very similar to C/C++, Java or C#.

Neural Lab Features

In version 3.1, the tutorial included with Neural Lab 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 try to incorporate background information on artificial neural networks.
One of the main features of Neural Lab is that it provides a visual environment to design and test artificial neural networks.
The tools allow reviewing and analyzing the structure of the training set.
In both versions, it is possible to see the activation of the neurons for each case in the data set. The tutorial of Neural Lab provides some examples in: prediction, data mapping, data classification and auto associative memory problems. Version 4.0 incorporates Kohonen networks that can be trained without supervision. The authors added in version 4.0 probabilistic neural networks.

Specific examples of neural networks include:

Artificial Neural Network for Prediction
Artificial Neural Network for Mapping
Artificial Neural Network for Auto Association
Artificial Neural Network for Classification
Artificial Neural Network Simulation

Neural Lab Uses Wintempla

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 the native language C++ and the native APIs of Win32. Once an artificial neural network has been trained, it is possible to save the network to a file. Then, the file can be open using Microsoft Visual Studio to create a standalone application that can access the trained artificial neural network.

Wintempla is a tool that integrates with Microsoft Visual Studio (students and professors may download a copy of Microsoft Visual studio from www.dreamspark.com. Wintempla provides a thin level of encapsulation to Win32 to simplify the creation of Web and Desktop applications using C++ and Object Oriented Programming. Wintempla main files are: Wintempla.h, Wintempla.cpp, WintemplaWin.h and WintemplaWin.cpp. One advantage of Wintempla is that the programmer has the option to use the native Win32 APIs or the Wintempla classes.

Wintempla includes

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 (remise, 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

File extensions

.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

References

Signal and Image Processing with Neural Networks by Timothy Masters.
Advanced Algorithms for Neural Networks by Timothy Masters.

External links

Neural Lab page http://www.fimee.ugto.mx/profesores/sledesma/documentos/
Download https://dl.dropboxusercontent.com/u/74744282/NeuralLab4_0.msi
Categories: Neural network software

Template:Uncut