Learning Sinkhorn divergences for supervised change point detection
This addresses the issue of poor detection in complex sequences for applications needing targeted change identification, though it is incremental as it builds on existing unsupervised approaches.
The paper tackles the problem of detecting change points in sequential data by introducing a supervised framework that learns a ground metric from labeled change points, enabling the use of Sinkhorn divergences for online detection. Experiments show substantial performance improvements over unsupervised methods with few labeled instances.
Many modern applications require detecting change points in complex sequential data. Most existing methods for change point detection are unsupervised and, as a consequence, lack any information regarding what kind of changes we want to detect or if some kinds of changes are safe to ignore. This often results in poor change detection performance. We present a novel change point detection framework that uses true change point instances as supervision for learning a ground metric such that Sinkhorn divergences can be then used in two-sample tests on sliding windows to detect change points in an online manner. Our method can be used to learn a sparse metric which can be useful for both feature selection and interpretation in high-dimensional change point detection settings. Experiments on simulated as well as real world sequences show that our proposed method can substantially improve change point detection performance over existing unsupervised change point detection methods using only few labeled change point instances.