Information Redundancy as an Information Theoretic Criterion for Onset Detection
Gert Van Dijck, Marc M. Van Hulle · 2006
We propose a criterion, called ‘maximal redundancy’, for onset detection in time series. The concept redundancy is inspired from information theory and indicates how well a signal locally can be explained by an underlying model exploiting past observations. It is shown that a local maximum in the redundancy is a good indicator for an onset. It is proven that ‘maximal redundancy ’ detection is a statistical asymptotically optimal detector for AR processes. Moreover, the detector accounts for non-Gaussianity of the innovations in the AR processes, using negentropy. Several applications are shown where the new criterion has been successfully applied.