- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper proposes CARMA, a variational autoencoder for multivariate time-series anomaly detection that combines timestep-level temporal latents with a window-level correlation latent. The latter is supervised to reconstruct the input window’s pairwise Pearson correlations using a Huber-plus-cosine objective and is broadcast into the signal decoder. At inference, anomalies are scored exclusively through signal reconstruction error, followed by temporal smoothing.
CARMA is coupling an explicitly supervised correlation representation to signal reconstruction. Evaluation on TSB-AD-M comprises 180 series from 17 datasets, 29 baselines, and three training seeds. CARMA achieves the highest reported mean VUS-PR, 0.4929, compared with 0.4409 for the strongest baseline. Additional evidence includes synthetic correlation-breaking anomalies, architectural and objective ablations, and paired statistical comparisons.
Critical strengths
Broad empirical improvement. Results cover 180 series and improve several complementary metrics, with repeated runs and statistical comparisons.
Useful supervision ablation. Adding correlation supervision improves VUS-PR from 0.4719 to 0.4929 within the same dual-latent architecture, supporting its value beyond additional latent capacity.
Critical weaknesses
Nonlinear dependency modeling is insufficiently validated. The explicit correlation target captures only linear dependence. The paper does not establish whether the correlation branch improves detection when nonlinear dependencies change while Pearson correlations remain unchanged, leaving its broader dependency-modeling claims insufficiently supported.
Computational efficiency is not evaluated. The paper provides no complexity analysis, FLOPs, inference latency, or throughput measurements. Consequently, the additional computational cost, accuracy–efficiency trade-off, and scalability to larger channel counts or windows remain unclear.
Questions for authors
Does correlation supervision improve detection of nonlinear dependency anomalies that preserve Pearson correlations? A comparison with the variant without correlation supervision would help establish the method’s scope.
What are CARMA’s computational complexity, FLOPs, inference latency, and throughput relative to the single-latent variant under identical hardware and input settings? Please distinguish training-only correlation computations from inference costs and clarify scaling with channel count and window length.
3 comments
W2 and Q2: What are CARMA’s computational complexity, FLOPs, inference latency, and throughput relative to the single-latent variant under identical hardware and input settings? Please distinguish training-only correlation computations from inference costs and clarify scaling with channel count and window length.
We provide a scaling cost analysis in the table below. The evaluated channel counts are 20, 50, 100 and the window lengths are 100, 150, and 200 respectively. CARMA’s FLOPs stay within a few percent of the single-latent model (no ) on every row. Computation increases with the length of the window. All evaluations are performed on the same hardware NVIDIA L40S 48GB VRAM, AMD EPYC 9454 48-Core Processor CPU, 128GB RAM. The training-only correlation computation time ms/window is 0.0018 (C=50, T=100)
Table. Scaling cost. Here SL = Single latent (no ) version, C=channel count, T=window length.
| Setting | CARMA params () | SL params () | CARMA FLOPs () | SL FLOPs () | CARMA latency (ms) | SL latency (ms) | CARMA throughput (/s) | SL throughput (/s) |
|---|---|---|---|---|---|---|---|---|
| , | 7.61 | 2.47 | 508 | 492 | 0.0169 | 0.0125 | 59.0 | 80.1 |
| , | 10.17 | 2.47 | 762 | 739 | 0.0229 | 0.0197 | 43.7 | 50.8 |
| , | 12.73 | 2.47 | 1016 | 985 | 0.0318 | 0.0276 | 31.4 | 36.3 |
| , | 7.78 | 2.50 | 515 | 500 | 0.0173 | 0.0127 | 57.7 | 79.0 |
| , | 10.34 | 2.50 | 773 | 750 | 0.0232 | 0.0199 | 43.1 | 50.2 |
| , | 12.90 | 2.50 | 1030 | 999 | 0.0322 | 0.0279 | 31.0 | 35.8 |
| , | 9.76 | 2.56 | 531 | 512 | 0.0176 | 0.0128 | 56.8 | 78.3 |
| , | 12.32 | 2.56 | 795 | 768 | 0.0242 | 0.0203 | 41.4 | 49.2 |
| , | 14.88 | 2.56 | 1058 | 1023 | 0.0335 | 0.0284 | 29.8 | 35.2 |
W2 and Q2: What are CARMA’s computational complexity, FLOPs, inference latency, and throughput relative to the single-latent variant under identical hardware and input settings? Please distinguish training-only correlation computations from inference costs and clarify scaling with channel count and window length.
(Continue...)
Additionally, we compare the mean time (s) per series on the TSB-AD-M dataset for CARMA with baselines in the table below. The second-best baseline is HYDRA with a VUS-PR of 0.4409 and a mean time per series of 61.29 seconds. The third baseline, StreamVAE, takes longer than CARMA and achieves a VUS-PR of 0.4277. A figure comparing the mean VUS-PR vs mean Time (s) per series on TSB-AD-M is uploaded (https://anonymous.4open.science/r/CARMA-AE8F/rebuttal/VUS-PR_vs_Time_compare_carma_and_baseline.png).
Table. Mean VUS-PR and mean Time (s) per series on TSB-AD-M. The best result in each column is in bold.
| Method | VUS-PR | Time (s) |
|---|---|---|
| CARMA | 0.4929 | 111.59 |
| CARMA w/o | 0.4620 | 83.02 |
| HYDRA | 0.4409 | 61.29 |
| StreamVAE | 0.4277 | 170.88 |
| PaAno_PAI | 0.4049 | 4.59 |
| USAD | 0.4026 | 6.29 |
| TimesNet | 0.3910 | 22.67 |
| PatchTST | 0.3902 | 21.74 |
| xLSTMAD | 0.3895 | 163.42 |
| FITS | 0.3642 | 50.31 |
| MSCRED | 0.3622 | 10.56 |
| OFA | 0.3557 | 132.24 |
| MMPAD | 0.3532 | 333.98 |
| MTAD_GAT | 0.3506 | 25.15 |
| GDN | 0.3318 | 10.67 |
| OmniAnomaly | 0.3153 | 112.30 |
| CNN | 0.2960 | 5.76 |
| CST_GL | 0.2937 | 35.01 |
| KMeansAD | 0.2867 | 36.36 |
| AutoEncoder | 0.2835 | 13.44 |
| PCA | 0.2691 | 91.86 |
| TranAD | 0.2651 | 13.12 |
| OCSVM | 0.2628 | 18.52 |
| IForest | 0.2487 | 60.07 |
| RobustPCA | 0.2379 | 76.54 |
| EIF | 0.2052 | 85.76 |
| COPOD | 0.2015 | 0.42 |
| HBOS | 0.1856 | 0.23 |
| KNN | 0.1693 | 30.74 |
| LOF | 0.1310 | 30.68 |
| AnomalyTransformer | 0.0989 | 30.19 |
Thank you for your comments. We believe the complexity analysis provides additional support for CARMA. We will include it in the appendix of the manuscript.
We thank you for finding the work useful and especially for observing the improvement coming from the dual latent architecture.
The question is very interesting to see if there's any improvement in the detection of nonlinear dependency anomalies. We have applied Pinet's et al. analysis to the TSB-AD-M dataset to isolate series that break only nonlinear relationships, while keeping the Pearson correlations unchanged [1]. But we didn't find any series that break only nonlinear relationships. All the segments that break nonlinear relationships are also either breaking Pearson correlations or are univariate type of anomalies. A summary table of anomaly types is uploaded (https://anonymous.4open.science/r/CARMA-AE8F/rebuttal/summary_tsbadm.xlsx). As a result we couldn't isolate any series that break only nonlinear relationships.
[1] M. Pinet, J. Cumin, S. Berlemont, and D. Vaufreydaz, “Anomalies in Multivariate Time Series Benchmarks Are Mostly Univariate,” Jun. 15, 2026, arXiv: arXiv:2606.02670. doi: 10.48550/arXiv.2606.02670.