- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper investigates the issue of "global shortcut contamination" that can arise in end-to-end graph structure learning for multivariate time series forecasting. The authors propose a "projection-based decoupling" approach, which removes dominant shared components via projection prior to graph learning. They introduce two implementations—MeanProj and PCA—and use a "last-observation persistence" strategy to reconstruct the removed coordinates. Experimental results demonstrate that this projection technique can significantly reduce out-of-distribution (OOD) errors in certain models and scenarios; however, its effectiveness is highly dependent on the specific model architecture and the chosen projection direction.
Critical strengths
- The theoretical analysis centers on specific failure modes.
- The paper repeatedly emphasizes that the "learned adjacency" should be interpreted as a "predictive computation graph."
Critical weaknesses
- Eqs. (2)–(5) serve more as an "existence proof" rather than a general characterization of practical end-to-end (E2E) graph learners.
- The paper lacks a rigorous link between the risk analysis based on binary support and the actual differentiable graph generator; consequently, the theoretical constructs do not perfectly align with the empirical models.
- While the theoretical results demonstrate that "certain shortcut edges can lead to lower risk," they do not prove that practical, non-linear E2E training will necessarily—or with high probability—learn these specific edges.
- The "projection intervention" affects not only the graph learning process but also the forecasting task itself.
- The synthetic experiments are conducted on a small scale: , with a limited autoregressive (AR) order and a simple rank-one factor structure.
- The real-world experiments are limited to only two traffic forecasting datasets (METR-LA and PEMS-BAY). Although the title refers to "multivariate forecasting," the paper lacks validation on multivariate time series (MTS) from other domains, such as energy, weather, finance, or industrial applications.
- The real-world experiments aggregate results from only three random seeds, and the tables lack standard deviations (std), confidence intervals (CI), or significance tests. In some cases, the observed performance gains are marginal.
- I will update my final score based on the authors' rebuttal.
Questions for authors
Q1. How can it be demonstrated that the gains truly stem from "graph decontamination" rather than from the projection process itself?
Q2. How does this approach compare to common methods for removing common components?
Q3. Does the actual "traffic shift" truly correspond to the "loading rotation" described in Theorem 2?
Q4. Is the graph obtained after projection genuinely "more correct"?
Q5. What is the impact of this method on in-distribution (ID) performance and on other domains?
Thank you for the detailed assessment. We added controlled completion, structure, rank, rotation, graph-utility, and cross-domain evaluations, while clarifying what the theory does—and does not—guarantee.
1. Binary support vs. nonlinear training
Addressed in: Sec. 2; Appendices A.3, A.8–A.9, G.6.
Theorem 1 gives an exact population-risk gap and strictness conditions in a specified one-step model, not merely an existence example. Since the conditional mean is linear under the Gaussian assumptions, the same gap holds for potentially nonlinear predictors with the same observed-coordinate access (Corollary 1).
For an actual feasible graph-related parameter direction d, we prove
DϕR[d]=−2E⟨Y−F,DϕF[d]⟩
(Proposition 3), a conditional local descent statement, not a convergence guarantee.
JVP/autograd matched numerically. At the smallest step, train loss decreased in 4/5 seeds for both tested directions, but OOD loss increased in all 30 direction/step/seed tests. Shared embeddings also affect forecasting weights; these are not graph-only interventions or observed training trajectories.
Q1. Source of gains
Addressed in: Secs. 3.2 and 4.5; Appendices G.2–G.3.
In a five-seed AGCRN synthetic experiment (N=32 observed variables, one node each), Original, Original with matched persistence, and projected-hard have mean OOD MSE 1.34422, 1.09065, 1.06519, respectively.
Matched output replacement reproduces 90.9% of the Original-to-projected-hard gain arithmetically; yet projected-hard improves on matched completion by 0.02546±0.00181 (5/5 paired seeds). This remainder includes other training differences; it is not identified as a graph-only effect.
With graph source varied under common forecasting conditions, the MeanProj-derived graph gives a smaller synthetic OOD benefit of 0.00336±0.00164; traffic fixed-graph results are mixed.
Q2. Common-component methods
Addressed in: Sec. 5.
Mean removal/PCA are not claimed as new. Unlike factor-adjusted network approaches such as FNETS [1], we study why prediction-optimal graph access diverges from local support and quantify loading-shift risk.
We retain the forecasting backbone and use last-observation completion. No numerical superiority over FNETS is claimed.
Q3. Traffic shift vs. loading rotation
Addressed in: Secs. 2 and 4.5; Appendix G.5.
Thank you for this question. We have not identified traffic time-of-day/weekend shifts as loading-only rotations.
A separate controlled rotation experiment with constant Bayes MSE tests Theorem 2: train-population per-variable excess MSE follows
0.1265625sin2θ.
Full OLS excess rises from 0.00592 at 0∘ to 0.13314 at 90∘, while local-mask OLS stays near 0.0804, reversing their ordering at larger angles (five seeds).
Fixed MeanProj/PCA do not outperform Full OLS under these rotations, highlighting a limitation of fixed training projectors.
Q4. Structural correctness
Addressed in: Sec. 4.5; Appendix G.4.
With N=64 variables and two generating factors, five-seed structural AUPRC increases from 0.0601±0.0226 (Original) to 0.6278±0.0547 (PCA rank 2); OOD MSE decreases from 1.5612±0.0972 to 1.0818±0.0208.
Both metrics improve over rank-one PCA in 5/5 paired seeds. We score symmetrized learned adjacency against undirected generating support, not causal edges. The generator does not satisfy all exact-projection assumptions; this is empirical ranking evidence, not guaranteed graph recovery.
Q5. ID, other domains, and reliability
Addressed in: Secs. 4.4–4.5; Appendices B, G.1, G.4, G.7.
Rank-two PCA improves synthetic ID MSE 0.27818→0.26995 (5/5 seeds).
We added Electricity (321 variables, MTGNN, five seeds). Chronological held-out test MSE (in 106 units) is 3.025±0.626 Original, 2.948±0.180 MeanProj, 9.671±0.023 PCA.
MeanProj improves MSE in only 3/5 seeds and worsens MAE; PCA worsens MSE in 5/5. The original traffic results remain three seeds, with ID/SD in existing tables; new experiments use five and report paired differences.
Thus, neither universal cross-domain improvement nor statistical significance of marginal effects is claimed.
Reference: [1] Barigozzi, Cho, and Owens (2024). “FNETS: Factor-Adjusted Network Estimation and Forecasting for High-Dimensional Time Series.” Journal of Business & Economic Statistics, 42(3), 890–902.