Neural Network Trainer with Second Order Learning Algorithms

Although neural networks have been around for over 20 years, we still have difficulties training them. Training is often difficult and time consuming. The paper describes a software (NNT) developed for neural network training. In addition to the traditional Error Back Propagation (EBP) algorithm, several second order algorithms were implemented. These algorithms are modifications of the Levenberg Marquet algorithm and they are able to train arbitrarily connected feedforward neural networks. In most cases the training process is more than 100 times faster than EBP training. These algorithms can also find solutions for very difficult networks where the EBP algorithm fails.