- Rating
- 3(+1)
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
CVENet targets exogenous-aware time series forecasting. The authors argue that prior work has two shortcomings: (i) a single prescribed treatment of non-stationarity is applied before endogenous–exogenous interaction, and (ii) a single functional form (typically an MLP) is applied to every token. To address these, the paper proposes three modules. Complementary View Construction builds an observation view and an instance-normalized view of the history at multiple scales, each carried by its own encoder stack with dual-correlation attention. Horizon-Wise View Reconciliation combines the two view forecasts with a learned per-forecast-step weight. Token-Wise Complementary Experts (TCE) route each token with top-k selection independently within a Fourier/frequency expert family and an MLP/time expert family, mix the two outputs with a token-dependent sigmoid gate, and add an always-on shared expert implemented with a depthwise convolution. Training combines a prediction loss, per-view supervision, correlation-modeling auxiliary losses, and a load-balancing regularizer. Experiments cover twelve public datasets under two exogenous-availability regimes, plus EAST-POINT127, a tokamak electron-density dataset with 40 actuator signals.
Critical strengths
- A coherent motivation and unified architecture.
The paper identifies two meaningful challenges in time series forecasting with exogenous variables: choosing how to handle non-stationarity and adapting token transformations to heterogeneous temporal patterns. CVENet addresses both within a unified framework rather than treating them as entirely separate problems. The resulting design is intuitive and potentially useful.
- Broad experimental evaluation.
The evaluation covers twelve public datasets, two future-exogenous-variable availability regimes, multiple forecasting horizons, and a tokamak dataset. The inclusion of both conventional forecasting benchmarks and an application involving actuator signals and electron-density measurements is valuable.
- Practical relevance.
The framework explicitly addresses the situation in which future exogenous variables are unavailable. This is important in practical forecasting, where covariate forecasts may be unavailable or unreliable at deployment time. The paper therefore targets a meaningful problem rather than a narrowly defined architectural benchmark.
Critical weaknesses
- The methodological novelty is not sufficiently distinguished from prior work.
CVENet combines several established ideas: multi-scale trend-seasonal decomposition, instance normalization, dual-correlation attention, mixture-of-experts routing, Fourier-based transformations, auxiliary supervision, and learnable forecast fusion. Although their integration may be useful, the paper does not convincingly establish which aspects constitute fundamental methodological advances rather than a composition of existing techniques.
- The experimental design does not fully isolate the sources of improvement.
CVENet introduces substantial architectural complexity. It maintains two encoder stacks, employs multiple expert networks and routers, adds a shared expert, and uses several auxiliary losses. The current ablations do not adequately disentangle the effects of increased parameter count, additional computation, independent view-specific parameters, and the proposed inductive biases.
- Statistical reliability of the reported improvements is unclear.
The paper reports competitive average performance, including reductions of 3.9% in MSE and 3.0% in MAE relative to DAG in the setting with future exogenous variables. These results are promising, but the aggregate improvements are not large enough to dismiss the possibility of run-to-run variation without further evidence.
- The proposed mechanisms are not sufficiently validated.
The paper motivates frequency-domain experts through their periodic inductive bias and horizon-specific reconciliation through the changing relevance of observations and standardized views. However, the reported results primarily establish predictive performance, rather than directly validating these explanations.
- Evaluation under missing future exogenous variables raises questions about error propagation.
When future exogenous variables are unavailable, CVENet uses predictions from its temporal branch as inputs to the channel branch. This is a reasonable design, but it introduces a potential error-propagation mechanism: errors in the predicted covariates may affect the final endogenous forecasts.
- Efficiency and reproducibility require stronger evidence.
The manuscript reports training on a single NVIDIA RTX 4090 and states that source code will be released after acceptance. However, the paper does not sufficiently establish the relative training cost, inference latency, memory consumption, or parameter efficiency of CVENet compared with strong baselines. These measurements are important because the proposed architecture duplicates view-specific processing and introduces multiple expert families.
Questions for authors
Refer to weaknesses.
5 comments
I thank the authors for the detailed response and for the substantial additional experiments. I acknowledge that most of my empirical concerns have been addressed or substantially mitigated.
However, my primary concern remains the methodological novelty (W1), and the response does not fully resolve it. The response clarifies the composition and demonstrates its effectiveness, but does not establish a fundamental methodological advance beyond a well-motivated arrangement of existing techniques. The empirical evidence supports effectiveness, not novelty.
To reconsider my score, I would need some theoretical support for the methodological design, even under simplifying assumptions (e.g., linear models or controlled synthetic settings). Specifically, a focused analysis showing why the two complementary views together strictly expand the representable function class, or reduce the achievable error, relative to either view alone would directly address my concern.
Follow-up Response to Reviewer jAWq
Dear Reviewer jAWq,
Thank you for your follow-up and for noting that most of your empirical concerns are resolved. We agree on W1: empirical gains show effectiveness, while novelty requires proving why the coupled design is necessary.
1. Structural Dilemma in TSF-X
Across power, weather, hydro, and plasma control (Section 4.1; Appendix A.1), single designs hit three limits:
- Level relaxation vs. -scaled gain. Over step , memory of fades while covariate gain scales with (grid clearing, hydro storage, or plasma transport lag; Wan et al., 2026). Observation views re-weight under fixed slope; standardization (Kim et al., 2022) captures -scaled gain yet locks , adding drift.
- Coupled views vs. naive stacking. Duplicating a view stays in one subspace, while unanchored fusion () leaves branches unidentifiable, requiring view loss and step-wise weights.
- Cycles plus jumps. Attention mixes cycles and covariate steps in one token stream (Figures 5–6), requiring token-wise Fourier and MLP routing.
2. Function-Class and Achievable-Risk Analysis
For () with mean , std (), , and , linear predictors (Sections 3.3, 3.5) form:
Proposition 1 (Function-Class Expansion). Over zero-mean shapes () for , () adds with (, ), while (affine ) locks and adds (). Homogeneous stacks collapse (), while for reconciling adds ():
Proposition 2 (Risk Reduction and ). Let (, , ; independent , of variance , zero-mean of variances ), , and . Single-view MSE floors are on and on (and nonlinear , as strips ). hits , yet leaves unidentifiable; under (Eqs. 11, 14), optimal have orthogonal errors and , so (Eq. 10) uniquely gives:
and shared (Table 3c) adds when varies with .
Dual-family experts (Eqs. 5–9). An -term Fourier sum fits tones yet has Gibbs MSE on jumps; a GELU MLP fits steps/ramps in units without periodic bias (Dong et al., 2025), motivating gate (Eq. 8).
3. Synthetic Check and Empirical Alignment
Over 5 seeds (2021–2025; , , , , , , ; 50k/50k), OLS confirms both:
| Variant | Theory | 5-Seed MSE |
|---|---|---|
| Observation view () | 0.383 | 0.383 ± 0.001 |
| Standardized view () | 0.536 | 0.538 ± 0.001 |
| Two observation views | 0.383 | 0.384 ± 0.001 |
| Two standardized views | 0.536 | 0.539 ± 0.001 |
| Shared | 0.248 | 0.248 ± 0.001 |
| Step-wise | 0.226 | 0.227 ± 0.001 (-40.9%) |
Learned rises from 0.029 () to 0.716 (), matching (, ). This ratio explains Figure 3—NP (, rising ), Sdwpfm1 (–, ), FR (, rising )—plus homogeneous (+8.7%–11.7%), widened single-view (+30.5%; Figure 4, Table 9), (+11.1%; Figure 11a, Table 9), and single-family switching (+40.4%–46.2%; Section 4.3) losses, and EAST-POINT127 horizon gains (96 ms: 0.024/0.026 720 ms: 0.289/0.331; Table 2).
We will add a summary in Section 3.5 and full proofs in Appendix C. Thank you again for your feedback.
Response to Reviewer jAWq
Dear Reviewer jAWq,
Thank you for your detailed review. Below we address all six points (additions marked in blue in the PDF).
W1: TSF-X Formulation vs. Existing Primitives
As noted in General Response Point 1 (Sections 1, 2.2, 2.3, 3.2), TSF-X couples two choices treated separately in prior work: keeping observation and standardized views in separate stacks (unlike single-normalization Autoformer, FEDformer, TimeMixer, FreTS, and FITS) and mixing parallel Fourier and MLP families per post-attention token (unlike homogeneous MLPs in DUET or covariate-free single-family layers in MoFE-Time). Across all 48 public benchmarks via Algorithm 1 (Appendix B.6, Table 10), FreTS (0.454), TimeMixer (0.476), FEDformer (0.563), and Autoformer (0.630) incur 42.1%–97.3% higher MSE than CVENet (0.319, a 29.6%–49.3% reduction) with future covariates, where CVENet ranks 1st across all 15 models; without future covariates (0.467 vs. 0.533–0.868), CVENet beats all four in dataset-averaged MSE and MAE on all 12 datasets.
W2: Isolating Capacity, Encoders, Experts, and Losses
Across all 24 public benchmarks (Section 4.3, Figure 4; Appendix B.5, Table 9), we tested matched-scale homogeneous controls and widened d, FFN width, or depth to match CVENet's size:
Sharing one encoder stack across both views raises MSE by 10.2% (24/24 benchmarks), widening DAG raises its own MSE from 0.332 to 0.383 (+15.2%, losing 22/24), and setting β=0, γ=0, or λ=0 raises MSE to 0.368 (+15.3%, 22/24), 0.355 (+11.1%, 23/24), and 0.339 (+6.2%, 16/24).
W3: Statistical Reliability Across Five Seeds
Over five paired seeds (2021–2025) on identical hardware and loaders across all 24 benchmarks, CVENet beats DAG in mean MSE on 19/24 benchmarks (sign test p=0.0033, winning 79/120 paired runs) and in mean MAE on 16/24 with 2 ties (p=0.026), alongside 20/24 wins in Table 7 (p=0.0008), with tight dispersion (e.g., PJM: 0.055±0.001 vs. 0.057±0.001 at F=24 and 0.117±0.009 vs. 0.130±0.005 at F=360; Sdwpfm1: 0.326±0.013 vs. 0.363±0.010 and 0.496±0.019 vs. 0.523±0.017; Colbún: 0.056±0.004 vs. 0.061±0.001 and 0.144±0.013 vs. 0.159±0.035). On 51-variable EAST-POINT127 (Table 2), CVENet wins 19/20 entries, cutting MSE by 14.3% over DAG and 27.7%–59.7% over TimeXer, TiDE, and TFT.
W4: Direct Validation of g(x) and ωh
W5 & W6: Error Propagation, Covariate Noise, and Efficiency
We are happy to discuss any of these results further.