Category Intelligent Software>Neural Network Systems/Tools

Abstract NeuroFusion is a general-purpose neural networks (NNs) library that can be used to create, train and apply constructive neural networks for solving both regression and classification problems. All theoretical information is hidden inside the NN library. You do Not have to do any tweaking using different training parameters or experiment with different architectures, activation functions, stopping conditions, and so on. Your development time is reduced significantly due to the fact that you have to deal with a minimum set of functions. Products features/capabilities include:

NeuroFusion library has 2 editions. The classic edition is written in American National Standards Institute (ANSI) C++ and compiled with Visual C++ 6.0. It can be used with any development environment that supports Windows dynamic-link library’s (DLLs). The .NET edition is written in C# and compiled using Visual Studio 7.1. Product can be used with any development environment that supports .NET technology.

NeuroFusion is Not an old-fashioned back-propagation product. This is a set of state-of-the-art constructive neural networks, combined with proprietary Alyuda algorithms for automatic data preprocessing, network design and training. These algorithms and heuristics are the result of years of research. They are successfully used in Alyuda's products and have proven to be efficient for a number of real-world applications.

The library does Not require any knowledge in neural networks or statistics. You just need to feed your data, call the training function and read the results. NeuroFusion makes all the mathematical calculations automatically. It preprocesses and partitions your data, initializes neural networks and even intelligently selects the stopping conditions and controls the generalization loss to provide you with the best solution neural networks can produce.

If you have some knowledge in neural networks, you can control the training process using real-time training statistics, modifying the stopping conditions, and specifying the data partition parameters. Manufacturer offers free technical support so you can get help and save time.

Unlike some other libraries, you can explore the functionality and performance of NeuroFusion due to a similar DLL used in Alyuda Forecaster XL (an additional product). You can download a free 30-day trial version of Forecaster XL and test the results you receive from NeuroFusion on your data. You can also request a library description, containing a help file and examples.

Products key features/capabilities include --

1) State-of-the art constructive neural networks;

2) Proven algorithms automatically create and train a network for you;

3) Extremely easy interface: you can run your first project within minutes;

4) Almost No artificial intelligence (AI) knowledge required: No need to tweak with parameters;

5) Dataset size and network size are limited only by available computer memory;

6) Automatic network adjustment for regression or classification problems;

7) Rigorously tested on many real-world problems.

Products additional features/capabilities include:

Data Preprocessing --

1) Input data types: numerical and categorical;

2) Automatic numerical columns scaling;

3) Automatic categorical columns encoding.

Training --

1) Automatic problem type determination (classification or regression);

2) Automatic network design and training;

3) Automatic and manual control of training stopping conditions;

4) Generalization loss control;

5) Full control of the training process, including pausing and stopping;

6) Real-time information about training progress;

7) 3 types of network errors: mean absolute error (MAE), mean square error (MSE), and correct classification rate (CCR).

Network Application --

1) The ability to save and load a network at any time that you need to;

2) Use a just created or loaded network to apply it to new data and get a neural network reply.

General --

1) Low resource requirements;

2) Example programs included in source form;

3) Support of load and save operations;

4) The Online Manual includes an "Introduction in Neural Networks" chapter.

Note: See G6G Abstract Number 20110> for additional product info from this manufacturer.

System Requirements


Manufacturer Web Site NeuroFusion

Price Real Time $499.00; Integration $1,999.00.

G6G Abstract Number 20111

G6G Manufacturer Number 100225