Neural Network Designer is a visual editor for designing networks based on the artificial neuron, the Perceptron. The application renders all design elements on a continuous basis, and can be used to test various design ideas on theoretical networks to see how the networks will perform.
Designing neural networks with Neural Network Designer is as simple as point-and -click. The networks are calculated and rendered automatically when given input. The outputs can be charted to oscilloscope.
Neural Network Designer is a work in progress.
One of the challenges ahead for Neural Network Designer is to mix back-propagation training with recurrent, or looped, network connectivity, and to introduce multidimensional arrays of perceptrons.
Currently, simple networks can be assembled with Neural Network Designer into arbitrary configurations, while the network is also simultaneously iterated from its input data.
Network learning is accomplished by input-output, point-to-point back-propogation.
