TigaByte Logo

Neural Networks and HyperNext


Whether you want to simply experiment with neural networks, develop your own or use the HyperNext neural networks for analysis and prediction then you will find that HyperNext is a great way to start.


For Analysis

HyperNext Creator now combines the functionality of Neural Creator and is a full visually based neural network development system that can be trained to perform pattern matching, feature recognition or make predictions on noisy or fuzzy data. It has an easy to use interface and comes with example projects and prebuilt neural network plugins allowing you to start training and testing straight away.

  • Of data in sales, marketing, investments, property, betting.
  • ...sports, medicine, science and engineering.
  • Perform pattern matching and recognition on noisy and fuzzy data.
For Study

Study at home with your own personal neural network system and perform practical experiments that help you fully understand how easy neural networks can be.

  • Study neural networks through practical experiments.
  • Design and evaluate your own neural networks.
  • Observe a neural network's response graphically.
  • Combine several neural networks to give advanced behaviour.
For Profit

The neural network technology provided with HyperNext Creator and Developer can allow you to develop your own neural network applications, whether with readily available neural network algorithms or perhaps using your own design.

  • Create your own standalone neural network applications.
  • Distribute your neural networks as HyperNext Player stacks.
  • Observe a neural network's response graphically.
  • Use Developer to build neural network plugins for HyperNext Creator.
What is HyperNext with Neural Networks?

HyperNext Creator is a flexible general purpose neural network development system that can be programmed to build standalone neural network systems and other types of software.

Neural networks can perform pattern matching, feature recognition or predictions on noisy or fuzzy data.

HyperNext Creator has a friendly user interface and a set of powerful neural network plugins that are "self programming". You act as teacher for a neural network by providing it with data and letting it know the goals it should learn. The neural network can then train itself using the data and goals provided, and during training can provide feedback on how well it is doing. Once you are satisfied that it has trained sufficiently well then it is ready to make a prediction about some new data. Assuming that the new data is derived from the same or similar sources as the training data then the neural network will be able to recognize features consistent with its past learning and advise you on its evaluation or prediction.

What can neural networks do for me?

There are a wide range of problems that can be solved using neural networks. Typical problems range from investment analysis, gambling, property analysis through to image and speech recognition. New applications for neural networks are being found all the time and you just need some inventiveness to see if your problem can be solved using this approach.

HyperNext Creator is a self contained system and unlike some other neural network packages does not need to use Microsoft Excel or other spreadsheet/graphing applications. It is flexible and sufficiently reprogrammable to be able to read in files and export its evaluations.

If you need to analyse some data then perhaps one of the accompanying neural network projects is suitable and need no customising. In which case an analysis involves specifying the input and output data and the neural network parameters after which training and testing can take place. If the input data has an inappropriate format then perhaps it can be converted using one of Creator's format commands otherwise it will be necessary to use its inbuilt programming languages to preprocess the data.

HyperNext Creator is ideal for experimenting and learning about neural networks because it comes with working projects that allow one to explore how a neural network responds as its parameters are changed. The output responses are displayed both graphically and textually so making it an ideal way to explore neural networks. For more advanced users HyperNext Creator's two programming languages allow user defined neural networks to be implemented and evaluated.

If a new neural network system needs developing then HyperNext Creator can easily handle this. It has two powerful programming languages available, namely HyperNext and RBscript. HyperNext is an easy to learn English-like programming language that is ideal for general tasks such as preprocessing data, reading/writing files, displaying graphics and for controlling neural networks. RBscript is a high speed object orientated language appropriate for implementing computationally intensive tasks such as neural networks. Together, these two languages make HyperNext Creator a powerful and flexible development tool.

Sometimes a neural network system needs to be implemented as a standalone application. HyperNext Creator can convert a project into a standalone application and also let the developer customise the look and feel of the application.

HyperNext Creator can also convert a project into a stack capable of running on the freeware HyperNext Player. Stacks are quite small in size and so more easily distributed by email than standalones.

How do I use HyperNext Neural Networks?

Basically there are two stages to using the neural network features of HyperNext Creator. During the first stage a working neural network capable of solving your particular problem is built but this stage can be skipped if one of the predefined projects meets your needs. If non of the projects are suitable then HyperNext Creator can be used to create a new project tailored specifically to your problem. If non of the neural network plugins are suitable then it is possible to program your own neural network using Creator's two inbuilt programming languages - HyperNext and RBscript. Once your neural network is working the next stage is to specify input data, output goals and neural network parameters. The neural network can then be trained on this data and then tested using a second batch of data.

Often a major problem with any intended data analysis or prediction regards the format of the required data. If your data is in a simple format such as a TAB delimited text file then it is an easy matter to read it in. However, if it has a binary format then you will need to understand the format so that HyperNext Creator can be programmed to read it.

Once you have the data in a usable format then HyperNext Creator can import the data, preprocess it, pass it to a neural network, and then carry out an evaluation. The results of the analysis can be saved to a file, displayed graphically or passed on to yet another neural network. The choice of how to process and analyse the data is yours but HyperNext Creator is highly flexible and can meet most needs.

HyperNext Creator comes with several practical projects that act as a starting point allowing you to experiment with neural network training and testing. Once you are comfortable with the practicalities of using a neural network then perhaps you can simply tailor one of the preexisting projects for your own use. However, if you need to develop your own unique solution then you will find that HyperNext Creator's two programming languages and visually based design environment enable you to tackle a wide range of problems. If your customised neural network is to be used in a number of HyperNext Creator projects then it is advisable to build a library plugin using HyperNext Developer and simply let your projects import the neural network.

If one of the accompanying neural network projects is suitable and need no customising then it is a fairly simple matter to set up your analysis. Your data is placed into the input field, the neural network parameters are specified together with any goals or outputs. Then your neural network is trained after which it is shown the data to be analysed.

HyperNext's Neural Network features

Although HyperNext Creator is a flexible general purpose software development system it does have features that aid the development of neural networks.

  • Visually based neural network design and data analysis system.
  • Includes working data analysis projects to get you started.
  • Easy to train and test on data using prebuilt neural network plugins.
  • Capability to process/analyse data using multiple neural networks.
  • HyperNext language for controlling neural network modules.
  • Can build standalone neural network systems.
  • Built-in help.
Typical Usage

Most languages use typed variables, that is, a variable must be declared to be of a specific type such as integer, string, real, boolean etc. For example, the compiler needs to know whether a particular variable will hold a whole number, a decimal number, a string and so on. In HyperNext though, all variables are stored as strings and need no type declaration, they just need their name declaring and an indication of whether they are Global or Local.

  • Set up the neural network parameters such as
  • ... inputs/outputs, network size.
  • Input your data from file or copy/paste.
  • Train the neural network on your data.
  • ... when satisfactory, your neural network is ready.
  • Show the neural network your new data.
  • Examine its response via graphs and text output.
 

TigaByte Software 2014

 
 

Information