Prediction and change detection in sequential data for interactive applications
Jun Wei Zhou, Li Cheng, Walter F. Bischof · Griffith Research Online · 2008
We consider the problems of sequential prediction and change detection that arise often in interactive applications: A semi-automatic predictor is applied to a time-series and is ex-pected to make proper predictions and request new human in-put when change points are detected. Motivated by the Trans-ductive Support Vector Machines (Vapnik 1998), we propose an online framework that naturally addresses these problems in a unified manner. Our empirical study with a synthetic dataset and a road tracking dataset demonstrates the efficacy of the proposed approach.