Detecting changes in high-dimensional time series is difficult because itinvolves the comparison of probability densities that need to be estimated fromfinite samples. In this paper, we present the first feature extraction methodtailored to change point detection, which is based on an extended version ofStationary Subspace Analysis. We reduce the dimensionality of the data to themost non-stationary directions, which are most informative for detecting statechanges in the time series. In extensive simulations on synthetic data we showthat the accuracy of three change point detection algorithms is significantlyincreased by a prior feature extraction step. These findings are confirmed inan application to industrial fault monitoring.
展开▼
