Technology

Classical Neural Networking

neural network

*This software implements a classical backpropagation supervised learning algorithm. For an overview of BAI SANI (Sequentially Activated Neuronal Input) neural network currently being developed with GIA/NLP, see Sequentially Activated Neural Networking.

BAI classical Artificial Neural Network technology includes the following features

  • Object Orientated - arbitrary number of neurons and unlimited connectivity between neurons with minimum memory usage
  • Supports Experience datasets and objects for real time training in environments (input/output information)
  • Supports both reading and writing of networks to XML (to store neural networks in computer/human readable format)
  • Supports recursive subnetting
  • Currently fitted with a standard back propagation algorithm
  • Supports 3D vector graphics report generation - can output its structure and properties to a 3D vector graphics file format for display using an OpenGL/RayTrace Renderer
  • Supports export to vector graphics file (SVG, LDR)
  • Operating system independent development - ISO C++ (currently available on Linux EL5/EL6/EL7/EL8/UB14/UB16/UB20 and Microsoft Windows XP/7/8/10)

For an overview of the biological features introduced by BAI ANN, see the "Neural Networking Biological Features" demonstration. This technology is not currently being maintained and it is recommended to use a modern vectorised and hard accelerated implementation (e.g. TensorFlow/PyTorch).

Technology Name Support Source License Source Documentation
Artificial Neural Network (ANN) LinuxRed Hat LinuxUbuntu LinuxWin Contact BAI (view source online) Auto Documentation