JAVA Applets for teaching neural networks

Five Java applets were developed for use with an undergraduate course on neural networks at Christian Brothers University. The paper starts with an overview of the course and a list of the hands-on software utilized. Some Java considerations are discussed; despite its weak points, Java was chosen for ease of Web delivery and platform independency. Detailed descriptions of the five Java applets follow, including implementation specifics and brief overviews of the underlying algorithms and theory. The five applets are as follows: SINGLE, a simulation of a simple single-layer network; DELTA, a demonstration of the delta rule learning algorithm; HOPFIELD, an application of a Hopfield network towards pattern recognition; BAM, an application of a bidirectional associative memory network towards pattern recognition and association; and CUMSEL, a demonstration of the cumulative selection process. The applets are also posted on the Web for public usage at http://www.cbu.edu/-pong/ai/