A neural network prediction of solar cycle 23

We examine the use of feed forward neural networks in the long term (i.e., years ahead) prediction of sunspot number. First, we briefly review the his tory of the time series and also some previous attempts to predict it. We o utline our neural network method and discuss how the reliability of the dat a affects training. We conclude that earlier data should not be used to tra in neural networks that are intended to make predictions at the current epo ch. We then use this understanding of the data in training neural networks, testing many different configurations to see which provides the best 1-6 y ear ahead prediction accuracies. By looking at the distribution of residual s, an estimate of the uncertainty is placed on the best networks' predictio ns. According to our predictions of yearly sunspot number, the maximum of c ycle 23 will occur in the: year 2001 and will have an annual mean sunspot n umber of 130 with an uncertainty of +/-30-80% confidence. Finally, we discu ss our result in relation to others and comment on how neural networks may be used in future work.