A novel anomaly detection method for multivariate time series based on spatial-temporal graph learning
Bibliographic record
Abstract
Anomaly detection for multivariate time series is crucial in real-world applications, including industrial equipment monitoring and predictive maintenance, financial risk management, smart city and traffic management, as well as environmental monitoring. However, the multivariate time series data exhibit high dimensionality, complex spatiotemporal dependencies, and nonlinear interactions between variables, making anomaly detection in such data a significant challenge. Most existing methods struggle to effectively capture these complex nonlinear relationships or explicitly model the spatiotemporal dependencies in multivariate time series. To address these limitations, we propose a novel approach for multivariate time series anomaly detection, called long short-term memory, temporal convolution and graph convolution (LTG), which is based on spatial-temporal graph learning. LTG uses Correlation Learning (CL) layer to acquire pairwise correlations, generating a graph adjacency matrix. This matrix is then fed into a Spatiotemporal Graph Neural Network (STGNN), composed of a Long Short-Term Memory (LSTM) Network, a Temporal Convolution Network (TCN), and a Graph Convolution Network (GCN), to capture the rich temporal and spatial dependencies in multivariate time series and accurately model these relationships. Finally, a PCA-based anomaly scorer is employed to output anomaly scores. Experimental results show that on the WADI and SMD datasets, LTG outperforms the state-of-the-art multivariate anomaly detection framework correlation-aware spatial-temporal graph learning (CST-GL) in both overall detection and early detection performance. Notably, on the WADI dataset, LTG’s average early detection performance across 6 different delay-constrained time points exceeds CST-GL by 10.64 percentage points.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.003 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".