A0241
Title: Online kernel CUSUM for change-point detection
Authors: Yao Xie - Georgia Institute of Technology (United States) [presenting]
Song Wei - Georgia Institute of Technology (United States)
Abstract: A computationally efficient online kernel cumulative sum (CUSUM) method is presented for change-point detection that utilizes the maximum over a set of kernel statistics to account for the unknown change-point location. The approach exhibits increased sensitivity to small changes compared to existing kernel-based change-point detection methods, including scan-B statistic, corresponding to a non-parametric Shewhart chart-type procedure. Accurate analytic approximations are provided for two key performance metrics: the average run length (ARL) and expected detection delay (EDD), which enable establishing an optimal window length to be on the order of the logarithm of ARL to ensure minimal power loss relative to an oracle procedure with infinite memory. Moreover, a recursive calculation procedure is introduced for detection statistics to ensure constant computational and memory complexity, which is essential for online implementation. Through extensive experiments on both simulated and real data, the competitive performance of the method is demonstrated, and the theoretical results are validated.
