Two-dimensional extensions of cascade correlation networks

Dynamic neural network algorithms are used for automatic network design in order to avoid time consuming search for finding an appropriate network topology with trial and error methods. The Cascade Correlation Network is a constructive method for building network architectures automatically. We present a novel incremental cascade network architecture based on it. We also report on benchmarking results for the two-spiral problem and two real world problems. Compared with results from the original cascade correlation network, our method yields a better performance.