|
February 22, 2003 |
I am working on a templated C++ library for artificial neural networks and genetic algorithms
started in 1996. This library is fully inlined and uses advanced template techniques (like
functors, specialization with assembly etc) for maximum performance and efficiency. BPN, GRNN, PNN
and Hybrid (trained by genetic algorithms) neural networks will be fully implemented by using
a smart-but-simple-and-lite layered architecture (the power of C++ templates!). At genetic programing
field, the standard WOF GA will be implemented and also I will present an evolutionary algorithm
that uses symbols and covers to the solution very fast compared to WOF algorithm
Because my old library has alot of standard C++ overhead (I was inspired by TurboVison's architecture
- was the best OOP example those DOS days) I'm completly rewrite the code and this will take some time
before I will be able to publish the code.
|
|