Does a Meeting in Santa Fe Imply Chaos? (Data Set A)
Pages 22

In addition to global predictors, in which each prediction is influenced by the entire learning set, and local predictors, based on a local (one hopes relevant) subset of points, we shall introduce predictors that take advantage of different reconstructions at different times (i.e., in different regions of phase space). This approach can provide better predictions than any single “optimal” reconstruction.