Untitled

Abstract

Abstract.We study the classification problem that arises when two variables-one continuous (x), one discrete (s)-evolve jointly in time. We suppose that the vector x traces out a smooth multidimensional curve, to each point of which the variable s attaches a discrete label. The trace of s thus partitions the curve into different segments whose boundaries occur where s changes value. We consider how to learn the mapping between the trace of x and the trace of s from examples of segmented curves. Our approach is to model the conditional random process that generates segments of constant s along the curve of x. We suppose that the variable s evolves stochastically as a function of the arc length traversed by x. Since arc length does not depend on the rate at which a curve is traversed, this gives rise to a family of Markov processes whose predictions are invariant to nonlinear warpings (or reparameterizations) of time. We show how to estimate the parameters of these models-known as Markov processes on curves (MPCs)-from labeled and unlabeled data. We then apply these models to two problems in automatic speech recognition, where x are acoustic feature trajectories and s are phonetic alignments.