Answers are grounded in this paper and cite their sources.
Detecting anomalies is a crucial task across various fields, such as cybersecurity, medical diagnostics, and industrial processes. While significant advancements have been achieved in developing anomaly detection methods, evaluating their effectiveness remains a significant challenge due to the limitations of existing datasets. Many current datasets are constrained in terms of size and realism, which can result in overly optimistic assessments of model performance. To address these limitations, we propose a novel dataset specifically tailored for anomaly detection, meticulously designed to closely simulate real-world conditions. We propose a hybrid approach that combines unsupervised and supervised machine learning techniques for detecting anomalies in large-scale time-series datasets. This approach significantly reduces the manual annotation required, as only a minimal amount of data needs labeling, thus eliminating the need for complete dataset annotation. The proposed method has shown robust performance on both publicly available and proposed time-series datasets.
No takes yet. Share an insight, caveat, or question.
Pai et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: