Algorithm animation of Neural Networks David Jackson and Ian G Morton Department of Computer PO Box 147 Liverpool United L693BX Kingdom Science University of Liverpool d.jackson@csc.liv. Abstract The internal operation of an artificial neural network is so radically different from more conventional programmed algorithms that it requires careful explanation. In an attempt to assist students with an understanding of how neural nets work we have applied algorithm animation techniques. Using the XTANGO system, a sequence of animations has been developed that is intended to be used in conjunction with lecture material. It begins with a study of a single perception, then moves on to the more powerful and complex Hopfield model of neural behaviour. ac.uk the extra clutter of the algorithm to be studied or altered without graphics code. The sequence of animations is intended to be used in conjunction with lecture material, and begins with a depiction of the singleperceptron model of neural behaviour. Subsequent animations accompany a study of more complex multi-neuron networks. 2 The Perception model 1 Introduction One of the problems that arise in teaching students about artificial neural networks is choosing a suitable level at which to describe their structure and operation. Viewing
/lp/association-for-computing-machinery/algorithm-animation-of-neural-networks-bON1CSa8tj