<ul><li>Anomaly detection in multivariate time series is challenging due to heterogeneous subsequence anomalies.</li><li>CATCH is a new framework based on frequency patching to improve anomaly detection.</li><li>CATCH uses a Channel Fusion Module (CFM) to capture fine-grained frequency characteristics and channel correlations.</li><li>Extensive experiments show that CATCH achieves state-of-the-art performance in anomaly detection.</li></ul>

CATCH: Channel-Aware multivariate Time Series Anomaly Detection via Frequency Patching

Discover more