Helmholtz flow matching improves time series anomaly detection accuracy
HCFM: Helmholtz Component Flow Matching for Time Series Anomaly Detection
Machine LearningArtificial IntelligenceDatabases
Summary
Detecting unusual patterns in time-series data is important but challenging because normal and anomaly signals get mixed together. The authors propose a way to separate these signals by dividing the model into parts that handle different aspects of the data flow, making it easier to spot anomalies. Their approach makes the detection more precise and faster to run. They tested it on many datasets and it worked better than existing methods.
What this means in practice
- •For industrial monitoring teams: Detect anomalies in sensor time-series data more accurately and efficiently to improve fault detection in manufacturing and infrastructure.
- •For financial analysts: Identify unusual patterns in financial time-series data to support risk assessment and fraud detection with improved precision.
Authors
Shyam Sundar Saravanan, Tie Luo, Sanjay Madria
Abstract
Flow Matching (FM) is an efficient simulation-free approach for training flow-based generative models, making it a promising foundation for scalable time-series anomaly detection (TSAD). However, standard FM learns a monolithic vector field in which nominal transport and local density-changing behavior are entangled. This creates a central challenge for TSAD; local expansion and contraction carry crucial anomaly-sensitive information, yet isolating this signal requires expensive Jacobian-trace estimation over the entire field. We propose \emph{Helmholtz Component Flow Matching} (HCFM), an unsupervised TSAD framework that parameterizes the vector field prediction network with separate transport, compression, and residual heads. The transport head is constrained to be divergence-free, so divergence estimation is restricted to the compression head, while a lightweight residual head absorbs the remaining mismatch. This structured parameterization improves scalability and detection performance by separating volume-preserving nominal transport from density-changing compression, yielding cleaner anomaly signals from distinct deviations of nominal behavior. Experiments on large-scale univariate and multivariate TSAD benchmarks show that HCFM consistently improves detection accuracy while maintaining efficient inference compared with 16 baselines. The code is available at https://github.com/TrustAIoT/HCFM__time-series-anomaly/