Noise-contrastive Online Change Point Detection

About

We suggest a novel procedure for online change point detection. Our approach expands an idea of maximizing a discrepancy measure between points from pre-change and post-change distributions. This leads to flexible algorithms suitable for both parametric and nonparametric scenarios. We prove non-asymptotic bounds on the average running length of the procedure and its expected detection delay. The efficiency of the algorithm is illustrated with numerical experiments on synthetic and real-world data sets.

Nikita Puchkin, Artur Goldman, Konstantin Yakovlev, Valeriia Dzis, Uliana Vinogradova · 2022

Related benchmarks

Task	Dataset	Result
Change Point Detection	Room Occupancy (test)	False Alarm Rate1	9
Change Point Detection	CENSREC-1-C Clean Record	FA Rate0	9
Change Point Detection	CENSREC-1-C SNR 20	FA0	9
Change Point Detection	CENSREC-1-C SNR 15	FA0	9
Change Point Detection	Synthetic data Example 1	Detection Delay5.9	5
Change Point Detection	Synthetic data Example 1 (test)	FA Rate0	5
Change Point Detection	Synthetic data Example 2 (test)	FA0	5
Change Point Detection	Synthetic data Example 3 (test)	FA Rate0	5
Change Point Detection	Synthetic data Example 4 (test)	False Alarm Rate0	5
Human activity detection	WISDM	FA Rate1	5