Time series classification using the Volterra connectionist model and Bayes decision theory

The authors describe the development of a new technique for determining the weights of a Volterra connectionist model (VCM) applied to the classification of stationary time series. This involves assigning a classification index to each class of time series and developing expressions for the state condition probability density functions such that the Bayes risk can be expressed as a function of the weights. The optimal weight values then correspond to the minimum Bayes risk.<>