Researchers have developed an open-source benchmark system called MMPAD for time-series anomaly detection using Matrix Profile methods. The system enhances traditional approaches by incorporating multidimensional aggregation, efficient k-nearest-neighbor retrieval for repeated anomalies, and moving-average post-processing. This technical report details the implementation, hyperparameter settings, and benchmark results for both univariate and multivariate time series, aiming to provide a reproducible reference for future research in this area. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Provides a reproducible open-source benchmark for time-series anomaly detection, potentially improving future research and applications in this domain.
RANK_REASON This is a technical report detailing an open-source benchmark and implementation for time-series anomaly detection, which falls under research.