- Rating
- 3(+1)
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper considers the spatiotemporal imputation problem in traffic data. It introduces MoHL, a mixture of hypergraph Laplacians that learns group-wise relational structure directly from incomplete spatiotemporal data.
Critical strengths
MoHL discovers sensor groups from co-observed residual correlations, learns weights through mask-matched holdouts, and adapts regularization to local evidence. Its central contribution is a compact parameter-sharing scheme, with reported advantages under day-long outages.
Critical weaknesses
- The paper explicitly shows that its hypergraph penalty is equivalent to a graph penalty with shared edge weights. This raises a question about the group construction: would a simpler correlation-based graph perform similarly? A baseline with matched connectivity and a comparable number of learned parameters would help establish what the groups contribute.
- The experimental setting limits the conclusions. The main results use 100 high-degree sensors from a selected seven-day window, and the deep baselines are trained from scratch on that window. I would like to see whether the gains hold across full networks and different time windows, and how the comparison changes when deep models can use longer training histories.
- The shared-versus-free comparison is useful, but the same optimization budget may not suit both models. Learning a separate weight for every edge may require more training or stronger regularization. Without checking these factors, it is difficult to attribute the performance gap specifically to useful group structure.
- I am not fully convinced by the claim that sharing helps most when observations are scarce. In Table 9, its advantage does not shrink as the observation window grows, while Table 2 includes settings where free weights perform better. Please explain how these findings support the claim and report uncertainty estimates for the main comparisons.
Questions for authors
See Weakness
12 comments
Reply to Reviewer m9Ec (2/7)
W1. Connectivity and parameter-count controls
Response.
Thank you for proposing this control. We agree that connectivity and parameter count need separate checks. Most controls trail MoHL on flow data, but simpler graphs win in some settings. The differing gate configurations limit attribution to grouping alone.
What we did.
We used MoHL’s residual correlations, profile, road-graph prior, solver and learning procedure, and tested:
- Matched connectivity with one weight.
- A top-5 graph with five per-rank weights.
- Top-5 graphs with one shared weight or free edge weights.
Tables M2 and M2b cover 100 sensors and all sensors. The controls omit the gate while MoHL retains it, so they do not isolate coupling structure.
Evidence and discussion.
Table M2. Correlation-graph controls: control − MoHL, grouped by what they match (100 sensors).
| Pairs | Learned weights | PEMS-BAY Cell | PEMS-BAY Cluster | PEMS-BAY Block-288 | PEMS03 Cell | PEMS03 Cluster | PEMS03 Block-288 | PEMS04 Cell | PEMS04 Cluster | PEMS04 Block-288 | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| MoHL (MAE) | 600–960 | 5 | 0.699 | 1.241 | 2.650 | 9.39 | 11.17 | 13.85 | 19.51 | 22.96 | 31.83 |
| Matched connectivity | |||||||||||
| threshold graph with MoHL's pair count, one weight | 600–960 | 1 | +0.005 | −0.001 | +0.074 | +0.89 | +0.78 | +0.62 | +0.08 | +0.19 | +0.60 |
| all pairs of MoHL's groups, one weight | 600–960 | 1 | +0.004 | +0.004 | +0.086 | +0.96 | +0.84 | +0.64 | +0.08 | +0.13 | +0.34 |
| Matched number of weights | |||||||||||
| top-5 graph, one weight per rank | 340–410 | 5 | +0.001 | +0.015 | +0.041 | +0.25 | +0.24 | +0.22 | +0.06 | +0.13 | +0.28 |
| Fewer or more weights | |||||||||||
| top-5 graph, one shared weight | 340–410 | 1 | +0.003 | −0.001 | +0.068 | +0.70 | +0.48 | +0.17 | +0.07 | +0.16 | +0.08 |
| top-5 graph, free weight per edge | 340–410 | 340–410 | +0.020 | +0.119 | −0.024 | −0.08 | +0.03 | +0.57 | +0.08 | +0.64 | +0.26 |
Positive differences favour MoHL; negative differences favour the control.
Matched controls. One-weight matched-support graphs trail on flow data; the per-rank control trails in every 100-sensor column. Connectivity or parameter count alone therefore does not reproduce the reported configuration’s accuracy, although the gate difference limits attribution.
What grouping changes. Couplings accumulate across shared groups and share weights by order, pooling evidence while allowing different effective pair couplings. This remains a pairwise quadratic penalty.
Full networks. The per-rank control still trails on PEMS03 day-long gaps; the threshold graph leads on PEMS-BAY Block-288. Most comparisons favour MoHL, with the same attribution limit.
Table M2b. Correlation-graph controls on the full networks: control − MoHL, grouped by what they match (all sensors, test weeks).
| Pairs | Learned weights | PEMS-BAY Cell | PEMS-BAY Cluster | PEMS-BAY Block-288 | PEMS03 Cell | PEMS03 Cluster | PEMS03 Block-288 | PEMS04 Cell | PEMS04 Cluster | PEMS04 Block-288 | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| MoHL (MAE) | 2,030–3,370 | 5 | 0.880 | 1.497 | 2.773 | 11.46 | 14.05 | 18.97 | 17.00 | 20.58 | 27.15 |
| Matched connectivity | |||||||||||
| threshold graph with MoHL's pair count, one weight | 2,030–3,370 | 1 | +0.012 | +0.026 | −0.040 | +0.71 | +0.20 | +0.26 | +0.12 | +0.23 | +0.46 |
| all pairs of MoHL's groups, one weight | 2,030–3,370 | 1 | +0.011 | +0.021 | +0.051 | +0.67 | +0.16 | +0.85 | +0.08 | +0.12 | +0.86 |
| Matched number of weights | |||||||||||
| top-5 graph, one weight per rank | 1,190–1,380 | 5 | +0.002 | +0.009 | +0.041 | +0.10 | +0.04 | +0.73 | +0.05 | +0.05 | +0.33 |
| Fewer or more weights | |||||||||||
| top-5 graph, one shared weight | 1,190–1,380 | 1 | +0.010 | +0.012 | +0.045 | +0.51 | +0.06 | +0.64 | +0.06 | +0.11 | +0.44 |
| top-5 graph, free weight per edge | 1,190–1,380 | 1,190–1,380 | +0.058 | +0.110 | +0.049 | +0.82 | +0.04 | +0.62 | +0.13 | +0.51 | +0.02 |
Positive differences favour MoHL; negative differences favour the control.
Conclusion.
MoHL generally outperforms simpler correlation graphs, including controls matched in connectivity or parameter count. The five-weight per-rank control has higher MAE in every reported setting at both sensor scales, supporting MoHL’s compact, structured design.
Reply to Reviewer m9Ec (3/7)
W2. Full networks, observation windows and training history
Response.
Thank you for these suggestions. MoHL is designed to reconstruct an incomplete recording when a long, representative training archive is unavailable. It learns sensor groups and shared weights from the current observations, pooling limited evidence while allowing the relational structure to be re-estimated for each window. Shared weights keep the estimation problem compact, while re-estimating the groups makes the model specific to the current window. We agree that robustness to network size and window choice, and the benefit of adding historical data, should be assessed separately.
Design rationale.
- Learn from the available window. The daily profile and sensor groups are estimated from observed entries, and the shared weights are learned through self-supervision on observed holdouts.
- Pool scarce evidence. Sharing weights by group order reduces the estimation burden compared with learning each pairwise coupling independently.
- Adapt the structure to the window. Re-estimating the groups allows the relational structure to reflect current co-variation. This supports flexible use when historical records are unavailable, limited or no longer representative.
What we did.
- Network size: we evaluated every method on all sensors and the test weeks, using only the week being completed.
- Window robustness: we tested additional weeks and longer windows at both 100 sensors and on the full networks (Tables M4 and M4b).
- Access to history: we trained the deep models on months of observations and gave MoHL the same history for its daily profile and groups.
Evidence and discussion.
Full-dataset accuracy. Table M3 repeats the M1 evaluation on all sensors and the test weeks.
Table M3. Full-dataset evaluation: MAE on all sensors, averaged over the test weeks.
| Method | PEMS-BAY Cell | PEMS-BAY Cluster | PEMS-BAY Block-288 | PEMS03 Cell | PEMS03 Cluster | PEMS03 Block-288 | PEMS04 Cell | PEMS04 Cluster | PEMS04 Block-288 |
|---|---|---|---|---|---|---|---|---|---|
| kNN-spatial | 3.890 | 4.736 | 4.091 | 72.276 | 82.536 | 73.180 | 75.894 | 92.907 | 76.847 |
| HaLRTC | 1.263 | 1.993 | 3.507 | 10.978† | 13.397 | 72.171 | 18.038 | 23.205 | 106.606 |
| LETC | 2.409 | 2.712 | 4.949 | 31.816 | 47.821 | 82.026 | 48.419 | 64.663 | 116.571 |
| WDGTC | 2.286 | 2.679 | 13.990 | 16.889 | 17.778 | 68.535 | 23.442 | 26.569 | 104.503 |
| GRIN | 0.874† | 1.424† | 2.885† | 11.273 | 15.158 | 27.541 | 17.629† | 23.273 | 56.956 |
| SPIN | 0.937 | 1.447 | 3.299 | 12.085 | 14.401 | 24.390 | 18.003 | 21.535 | 35.038 |
| ImputeFormer | 1.071 | 1.474 | 3.061 | 12.356 | 13.949 | 20.385† | 17.931 | 20.946† | 29.662† |
| MoHL | 0.872 | 1.422 | 2.697 | 10.974 | 13.400† | 16.719 | 16.991 | 19.747 | 25.928 |
| Improvement (%) ↑ | +0.23 | +0.14 | +6.52 | +0.04 | −0.02 | +17.98 | +3.62 | +5.72 | +12.59 |
Bold: lowest MAE; †: second lowest. Positive improvement means lower MAE than the best baseline.
Performance in the intended setting. With only the current week available, MoHL has the lowest reported mean under day-long gaps on all three full networks. This extends the long-gap result beyond the original 100-sensor subset. Short-gap margins are smaller and rankings vary by regime, including HaLRTC’s slight lead on PEMS03 Cluster.
Different deployment scenarios. An established network may have months of representative records from which a history-trained model can learn recurring patterns; that model is more accurate in our history experiment. A new deployment or a short recording released for analysis may not provide such an archive, and older records may no longer reflect current sensor relationships. MoHL can be fitted from the observed entries of the current window and rebuild its groups for subsequent windows, so deployment does not depend on collecting a separate pretraining archive. The history experiment therefore evaluates the benefit of an additional data resource, while MoHL addresses flexible reconstruction from the data currently available.
Reply to Reviewer m9Ec (1/7)
Thank you for your thoughtful review, constructive suggestions, and patience. We apologize for the delay in responding. We took additional time to complete a substantial set of experiments so that we could address your concerns carefully and thoroughly.
We address your comments in the replies that follow. For reference, Table M1 reports the full-dataset evaluation on all sensors, averaged over the test weeks.
Table M1. Full-dataset evaluation: MAE on all sensors, averaged over the test weeks.
Mean ± s.d. across test weeks. Bold: lowest MAE; †: second lowest.