Simple but very fast C++ back propagation neural network library. Could be used for approximations, models identification, patterns recognition, multi-factor prognosis, etc
Contains:
-
TNeuroNet - neural network class with 3 learning functions
- Standard Back propagation error learing function
- SCG - scaled conjugate gradient learing function (fastest, smallest SSE)
- Learing function with genetic algorithm (unique results: SSE on test patterns are less than on learning pattern)
and 3 activation functions:
- Standard logistic activation function (best)
- Hyperbolic tangent activation function
- Exponential activation function
-
TGeneticAlgorithm - real value coding genetic algorithm
-
TDataSource - abstract class containing data source for neuro net learning
-
TLearnPattern - class contains normalized data patterns for neuro net learning
-
Some other helper classes