Neural Network Trainer through Computer Networks

This paper introduces a neural network training tool through computer networks. The following algorithms, such as neuron by neuron (NBN) , error back propagation (EBP), Levenberg Marquardt (LM) and its improved versions are implemented in two different computing methods, traditional forward-backward computation and newly developed forward-only computation. The training tool can handle not only conventional multilayer perceptron (MLP) networks, but also arbitrarily connected neuron (ACN) networks. There are several benefits that make this network training tool desirable. Network training process can be used remotely through any network connection or any operating system can be used to access the network training tool, making the application operating system independent. Also much less installation time and configuration time is required because the training tool locates on one central machine. Many users can access at the same time. Users can train and see the training results directly through networks. And the most important thing is the software producers can protect their intellectual property when network browsers are used as user interface.