Pith. sign in

REVIEW 4 major objections 5 minor 34 references

CARNet claims that injecting learned cycles into linear-complexity core aggregation outperforms attention-based multivariate forecasters on most benchmark settings.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 07:47 UTC pith:QNTJZ4PO

load-bearing objection CARNet is a legitimate architectural combination, but the paper's central SOTA claim is undercut by its own budget-controlled comparison; it deserves a serious referee, not a desk reject. the 4 major comments →

arxiv 2607.21681 v1 pith:QNTJZ4PO submitted 2026-07-23 cs.LG

CARNet Cycle-Conditioned Core Aggregation and Redistribution for Multivariate Time Series Forecasting

Awsaf Tausif Adib , Md. Shahria Sarker Shuvo , Md. Estehaar Ahmed Emon , Mustafa Kamal , Fuad Rahman , Shafin Rahman , Nabeel Mohammed This is my paper
classification cs.LG MSC 68T0762M10
keywords multivariate time series forecastingcross-variate dependenciesperiodic cyclescore aggregationattention-freelinear complexitylong-term forecastingchannel-dependent modeling
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

CARNet claims that explicit global cycle information can be injected into an attention-free, linear-complexity cross-variate aggregation model without sacrificing efficiency, and that doing so improves forecasting accuracy on real-world multivariate time series. To show this, the authors add a learnable recurrent cycle representation to both the aggregation and redistribution stages of a core-based architecture, and partition the core into multiple heads. On twelve standard benchmarks across four horizons each, they report the best MSE in 38 of 48 settings and best MAE in 42, with ablations attributing the largest gain to cycle-conditioned redistribution. If the comparison holds, periodicity-aware core aggregation could become a competitive alternative to attention-based channel interaction.

Core claim

The central discovery claimed is that conditioning cross-variate core interactions on a phase-aligned recurrent cycle, rather than simply subtracting the cycle as a residual, lets a linear-complexity, attention-free model capture periodic structure and inter-variable dependencies jointly. CARNet learns a cycle matrix per channel, aligns it with the input window, fuses it into variate embeddings, extracts a shared core through Multihead Core Aggregation by partitioning the feature dimension into h subspaces, pools across variates, and then redistributes the core together with the cycle projection back into each variate. The paper argues that removing any of these stages degrades performance,

What carries the argument

Multihead Core Aggregation (MHCA) with cycle-conditioned redistribution. MHCA splits the variate embedding into h heads, transforms each head with a grouped 1x1 convolution, merges head-wise cores, and pools across variates to form a global core; the learned recurrent cycle Z, phase-aligned by the estimated stable cycle length W via autocorrelation analysis, is injected both before aggregation and during redistribution. This carries the argument because it turns the global periodic pattern into a conditioning signal for cross-variate interaction while keeping per-layer complexity linear in both the number of variates and the sequence length.

Load-bearing premise

The comparison assumes that the baseline numbers taken from earlier papers were produced under the same training and evaluation protocol as CARNet's own runs, so the reported win counts reflect a fair race.

What would settle it

Rerun every baseline under CARNet's exact training recipe, including the same number of epochs, one-cycle learning-rate schedule, look-back length, and random seed, then count wins on the same 48 settings; if the 38-of-48 MSE margin shrinks or reverses, the claimed superiority is not established.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • On strongly periodic benchmarks such as Electricity and PEMS, CARNet reports the best error in a majority of settings, implying that cycle-conditioned core interaction can match or beat attention-based channel modeling.
  • Ablations indicate that cycle-conditioned redistribution is the most valuable component, so future core-based designs should inject periodicity at redistribution, not only at input decomposition.
  • The per-layer complexity is linear in both variates and time, so if the claims hold, the architecture scales to high-dimensional multivariate problems more cheaply than quadratic attention.
  • The cycle-length ablation shows milder degradation than a comparison model when W is misspecified, suggesting the mechanism tolerates imperfect periodicity estimates.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The cycle-conditioning idea is not inherently tied to this particular core design; a fair test would be porting the same conditioning mechanism into other linear channel-mixing architectures and measuring whether similar gains appear.
  • Because the paper fixes W at the dataset level, an extension to per-sample or adaptively estimated cycle lengths is natural, and the reported robustness to harmonic values of W suggests that such adaptivity could work without retuning.
  • The fixed-epoch and multi-seed results suggest the cycle prior accelerates convergence, but the paper does not directly convert this into an accuracy-per-training-epoch comparison; that trade-off is worth quantifying for deployment decisions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes CARNet, an attention-free multivariate time series forecasting model that extends SOFTS-style core aggregation with learnable cycle conditioning and a multi-head core aggregation mechanism. The authors claim that CARNet consistently outperforms strong transformer and non-attention baselines across 12 benchmarks with linear complexity. The paper reports MSE/MAE results, ablations, efficiency analyses, and a fixed-epoch comparison. The central contribution is empirical: the architecture is presented as a new state-of-the-art for periodic multivariate forecasting.

Significance. If the empirical claims were supported, CARNet would be a useful efficient alternative to attention-based channel-dependent models, and the combination of periodic conditioning with core aggregation is a plausible design direction. The complexity analysis and the ablation study provide useful evidence about the contribution of individual components. However, the core empirical claim is not established by the reported experiments, and the paper's own controlled experiments undermine the headline message.

major comments (4)
  1. [Table 2 / §4.2] The claimed 'best performance in 38 out of 48 settings' is not reproducible from the printed table. Examples where CARNet is not best: Traffic H=96 (SOFTS 0.376 vs CARNet 0.430), ETTh2 H=96 (TimeXer 0.286 vs CARNet 0.295), ETTm2 H=96 (CycleNet 0.163 vs CARNet 0.171). In fact, CARNet has no best MSE on ETTh2, ETTm2, or Traffic average. The count in §4.2 is therefore contradicted by the paper's own data.
  2. [§4.3, Table 4] The main comparisons in Table 2 are uncontrolled: CARNet was trained for 30 epochs (7 for ETTm), while baseline numbers were adopted from TQNet/SOFTS, with TimeXer and iTransformer originally trained for only 10 epochs. The authors' own fixed-budget experiment (Table 4) shows that at 10 epochs CARNet loses to TQNet on ETTm1 avg MSE (0.378 vs 0.377) and to TimeXer on Weather avg MSE (0.243 vs 0.241). This directly undercuts the claim that gains come from the architecture rather than from longer optimization.
  3. [§4.1, §4.3, Appendix F] No error bars or significance tests are provided for any baseline. Baseline results are adopted from prior papers rather than rerun under CARNet's training protocol. Appendix F reports seed variance only for CARNet, not for baselines. The claimed 'consistent outperformance' is therefore not statistically established, and the comparisons are not controlled.
  4. [Table 2, ECL row] The ECL average row appears corrupted: SOFTS avg MSE is printed as 0.270 and iTransformer as 0.365, while the individual horizon values for these models are around 0.14–0.22. This is likely a formatting/alignment error, but because the win count is derived from this table, the error further undermines the reliability of the headline result.
minor comments (5)
  1. [References and notation] TimeXer is cited as [4] in the text but appears as [19] in the Table 2 header. 'TQNet' is inconsistently rendered as 'TQnet' in Table 2. The '1st Count' row in Table 2 is a summary rather than a model column and should be clearly separated.
  2. [Algorithm 1, Eq. (2)] Algorithm 1 line 4 writes V' ← ψ(L_V(ψ(V))), whereas Eq. (2) defines V' = ψ(L_V(F_A(X',Z'))). The extra ψ in the pseudocode appears inconsistent and should be corrected.
  3. [Figure 5] The parameter counts listed in Figure 5 (124k, 1.04M, etc.) are not clearly mapped to model names in the caption. It is ambiguous which count corresponds to which model.
  4. [Figure 7 caption] 'produces predictions that are more accurate with the ground-truth series' should be worded as 'more aligned with the ground-truth series.'
  5. [Appendix B] Appendix C states a 6:2:2 train/validation/test split, but the split sizes in Table 5 (e.g., ETTh1 8545/2881/2881 of 14,400) correspond to approximately 12:4:4. The text and table should be reconciled.

Circularity Check

0 steps flagged

No circular dependency found; CARNet's empirical claims rest on external benchmarks rather than on its own assumptions.

full rationale

CARNet is an empirical architecture paper. The model equations (1)-(5) define a forward pass (cycle projection, attention-free core aggregation, redistribution, output projection) and the training loss (6) is standard MSE; no output quantity is defined in terms of the target it purports to predict, and no fitted parameter is renamed as a prediction. The cycle length W is taken from prior external work (CycleNet/TQNet) via ACF, and the cycle matrix is jointly trained, but the forecasting targets are future observations, not the cycle representation, so this is not circular. The main comparison adopts baseline numbers from TQNet and SOFTS (Table 2 caption), and Section 4.3 later introduces a fixed 10-epoch comparison; this is a comparison-protocol/fairness issue, not a circularity. There is no self-citation chain: all load-bearing references are to external authors (Lin et al., Zhang et al., Woo et al.), not to the present authors. No equation reduces to another by construction, and the complexity analysis is independent. The skeptical concern about mismatched training budgets and non-reproducible win counts is a correctness/evidence concern, which is outside the circularity definition and should not raise the circularity score.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

This is an empirical deep-learning paper. The main free parameters are dataset-level cycle lengths, per-dataset learning rates, and grid-searched architecture dimensions. No new physical entities are introduced. The axioms are standard domain assumptions for forecasting benchmarks, though the single-cycle-per-dataset assumption is explicitly acknowledged as limiting.

free parameters (3)
  • Cycle length W per dataset = ETTh1/2: 24; ETTm1/2: 96; ECL/Traffic: 168; Weather/Solar: 144; PEMS: 288
    Estimated from autocorrelation analysis (ACF) on each dataset (Section 3.2, Table 1) and used to phase-align the recurrent cycle matrix. Although presented as intrinsic, W is a data-derived number and the ablation in Figure 6 shows performance varies with it.
  • Learning rate per dataset = 1e-4 for ETT, Weather, Solar; 1e-3 for ECL, Traffic, PEMS
    Reported in Table 1 and not governed by a single protocol; selected per dataset, which affects the comparison.
  • Model hyperparameters (d, d', d_ff, e_layers, batch size, heads) = d, d_ff in {128,256,512}; d' <= d; layers {1,2,3,4}; batch {16,32}; head count not specified in main text
    Chosen by grid search (Section 4.1). These choices affect accuracy-efficiency trade-offs and are not fixed a priori.
axioms (3)
  • domain assumption Autocorrelation analysis identifies a stable dominant cycle length shared across all variates in a dataset
    Section 3.2 estimates W via ACF and assumes a single dataset-level cyclic representation Z is phase-aligned for every channel and every training sample.
  • domain assumption Mean squared error is the appropriate training and evaluation objective
    Section 3.3 and Section 4.1 use MSE for optimization and as the primary metric, following standard practice but biasing evaluation toward this loss.
  • domain assumption Instance normalization statistics computed over the look-back window generalize to the forecast horizon
    Section 3.4, Eq. 7 applies per-channel mean/standard deviation denormalization; this is the standard RevIN assumption that distribution shifts are captured by the look-back window.

pith-pipeline@v1.3.0-alltime-deepseek · 20519 in / 14760 out tokens · 129342 ms · 2026-08-01T07:47:34.706032+00:00 · methodology

0 comments
read the original abstract

Accurately modeling cross-variate dependencies remains a key challenge in multivariate time series forecasting, particularly in the presence of strong periodic patterns. Many existing approaches rely on attention-based mechanisms that incur quadratic complexity and scale poorly with increasing numbers of variates. Recent attention-free aggregation models address this issue through linear-complexity core-based interactions, but they do not explicitly leverage the global periodic structure present in the data. To overcome this limitation, we propose CARNet, a Cycle-Conditioned Core Aggregation and Redistribution framework that integrates global recurrent cycle information into efficient core based interaction modeling via Multihead Core Aggregation. Extensive experiments on multiple real-world multivariate forecasting benchmarks demonstrate that CARNet consistently outperforms strong transformer and non-attention baselines across diverse prediction horizons while preserving linear-complexity modeling of cross-variate dependencies.

Figures

Figures reproduced from arXiv: 2607.21681 by Awsaf Tausif Adib, Fuad Rahman, Md. Estehaar Ahmed Emon, Md. Shahria Sarker Shuvo, Mustafa Kamal, Nabeel Mohammed, Shafin Rahman.

Figure 1
Figure 1. Figure 1: Comparison of core aggregation strategies. (a) The traditional approach ag￾gregates all variates into a single shared core representation. (b) Our Multihead Core Aggregation first partitions variates into multiple interaction subspaces, extracts head￾specific cores, and then combines them into a shared core, enabling richer and more diverse cross-variate interaction modeling. 2.3 Channel-Dependent Modeling… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the CARNet framework. (a) In the Cycle-Conditioned Core Ag￾gregation module, the variate embedding and cycle projection are fused to obtain a cycle-conditioned variate embedding. (b) The Multihead Core Aggregation module ex￾tracts a compact core representation from the cycle-conditioned variate embedding. (c) In the Cycle-Conditioned Core Redistribution module, the shared core is redistributed … view at source ↗
Figure 3
Figure 3. Figure 3: Performance on different look-back length on ECL dataset. Cycle-Conditioned Integration vs. Residual Cycle Decomposition. To examine whether CARNet’s gains come only from recurrent cycle modeling, we perform an ablation by integrating Residual Cycle Forecasting (RCF) [16] into the vanilla SOFTS framework. Following this setup, SOFTS is applied to the residual series after removing the recurrent cycle compo… view at source ↗
Figure 4
Figure 4. Figure 4: Comparison between CARNet and SOFTS augmented with Residual Cycle Forecasting (RCF) across six datasets, averaged over four prediction horizons. As shown in [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Efficiency Analysis of CARNet On ECL Dataset. Cycle Length Ablation. To assess the sensitivity of CARNet to the choice of cycle length W, we evaluate the model across multiple values of W on the ECL dataset, averaged over four prediction horizons, while keeping all other hyperparameters fixed. The cycle lengths considered follow those reported in [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Effect of cycle length W on forecasting performance (averaged MSE and MAE over four prediction horizons) on the ECL dataset. CARNet is compared against TQNet [17] across the same set of cycle lengths, with W=168 corresponding to the dominant weekly periodicity. Evaluation Under Fixed Training Epoch. To ensure a fair comparison with recent state-of-the-art models such as TimeXer [4] and iTransformer [2], wh… view at source ↗
Figure 7
Figure 7. Figure 7: Visualization of forecasting results on the ECL dataset with look-back window 96 and prediction horizon 96. 5 Limitations In this work, we introduce CARNet, an attention-free approach for jointly mod￾eling recurrent cycles and cross-variate dependencies. Despite its effectiveness, CARNet has several limitations. First, similar to CycleNet [16] and TQNet [17], the cycle length W is determined at the dataset… view at source ↗
Figure 8
Figure 8. Figure 8: Visualization of forecasting results on the PEMS08 dataset with look-back window 96 and prediction horizon 96. (a)CARNet (b)TimeXer (c)TQNet 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 1 .7 1 .6 1 .5 1 .4 1 .3 Grou n dTru th Pre d iction 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 1 .7 1 .6 1 .5 1 .4 1 .3 Grou n dTru th Pre d iction 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 1 .7 1 .6 1 .5 1 .4 1 .3 Grou … view at source ↗
Figure 9
Figure 9. Figure 9: Visualization of forecasting results on the ETTm1 dataset with look-back win￾dow 96 and prediction horizon 96 [PITH_FULL_IMAGE:figures/full_fig_p021_9.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

34 extracted references · 4 linked inside Pith

  1. [1]

    In: Proc

    Nie, Y., Ma, Q., Zhang, Q., Xu, Y., Wang, X.: A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. In: Proc. Int. Conf. on Learning Rep- resentations (ICLR) (2023)

    2023

  2. [2]

    In: Proc

    Liu, H., Lin, Z., Zhang, Y.: iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. In: Proc. Int. Conf. on Learning Representations (ICLR) (2024)

    2024

  3. [3]

    arXiv preprint arXiv:1607.08022 (2016)

    Ulyanov, D., Vedaldi, A., Lempitsky, V.: Instance Normalization: The Missing Ingredient for Fast Stylization. arXiv preprint arXiv:1607.08022 (2016)

    Pith/arXiv arXiv 2016

  4. [4]

    In: Proc

    Woo, G., Kim, S., Song, H.O.: TimeXer: Learning Multi-scale Temporal Represen- tations for Long-term Forecasting. In: Proc. Int. Conf. on Learning Representations (ICLR) (2024)

    2024

  5. [5]

    et al.: BigBird: Transformers for Longer Sequences

    Zaheer, M. et al.: BigBird: Transformers for Longer Sequences. In: Proc. Advances in Neural Information Processing Systems (NeurIPS) (2020)

    2020

  6. [6]

    et al.: Informer: Beyond Efficient Transformer for Long Sequence Time- Series Forecasting

    Zhou, H. et al.: Informer: Beyond Efficient Transformer for Long Sequence Time- Series Forecasting. In: Proc. AAAI Conf. on Artificial Intelligence (AAAI) (2021)

    2021

  7. [7]

    In: Proc

    Gu, A., Dao, T.: Mamba: Selective State Space Models. In: Proc. Int. Conf. on Learning Representations (ICLR) (2024)

    2024

  8. [8]

    et al.: Is Mamba Effective for Time Series Forecasting? Neurocomputing (2025)

    Wang, Z. et al.: Is Mamba Effective for Time Series Forecasting? Neurocomputing (2025)

    2025

  9. [9]

    et al.: TimePro: Efficient Multivariate Long-term Time Series Forecast- ing with Variable- and Time-Aware Hyper-state

    Ma, X. et al.: TimePro: Efficient Multivariate Long-term Time Series Forecast- ing with Variable- and Time-Aware Hyper-state. In: Proc. Int. Conf. on Machine Learning (ICML) (2025)

    2025

  10. [10]

    et al.: Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting

    Wu, H. et al.: Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting. In: Proc. NeurIPS (2021)

    2021

  11. [11]

    et al.: FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting

    Zhou, T. et al.: FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting. In: Proc. ICML (2022)

    2022

  12. [12]

    et al.: TSMixer: An All-MLP Architecture for Time Series Forecasting

    Liu, Y. et al.: TSMixer: An All-MLP Architecture for Time Series Forecasting. arXiv preprint arXiv:2306.09364 (2023)

    Pith/arXiv arXiv 2023

  13. [13]

    et al.: SOFTS: Series-Core Fusion Transformer for Multivariate Time Series Forecasting

    Zhang, X. et al.: SOFTS: Series-Core Fusion Transformer for Multivariate Time Series Forecasting. In: Proc. NeurIPS (2024)

    2024

  14. [14]

    et al.: Graph WaveNet for Deep Spatial-Temporal Graph Modeling

    Wu, Z. et al.: Graph WaveNet for Deep Spatial-Temporal Graph Modeling. In: Proc. IJCAI (2019)

    2019

  15. [15]

    et al.: SparseTSF: Modeling Long-term Time Series Forecasting with 1k Parameters

    Lin, S. et al.: SparseTSF: Modeling Long-term Time Series Forecasting with 1k Parameters. In: Proc. ICML (2024)

    2024

  16. [16]

    et al.: CycleNet: Enhancing Time Series Forecasting through Modeling Periodic Patterns

    Lin, S. et al.: CycleNet: Enhancing Time Series Forecasting through Modeling Periodic Patterns. In: Proc. NeurIPS (2024)

    2024

  17. [17]

    et al.: Temporal Query Network for Efficient Multivariate Time Series Forecasting

    Lin, S. et al.: Temporal Query Network for Efficient Multivariate Time Series Forecasting. In: Proc. ICML (2025)

    2025

  18. [18]

    et al.: Are Transformers Effective for Time Series Forecasting? In: Proc

    Zeng, A. et al.: Are Transformers Effective for Time Series Forecasting? In: Proc. AAAI (2023)

    2023

  19. [19]

    et al.: TimeMixer: Decomposable Multiscale Mixing for Time Series Fore- casting

    Liu, Y. et al.: TimeMixer: Decomposable Multiscale Mixing for Time Series Fore- casting. In: Proc. ICLR (2024)

    2024

  20. [20]

    et al.: TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis

    Wu, H. et al.: TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis. In: Proc. AAAI (2023)

    2023

  21. [21]

    et al.: PyTorch: An Imperative Style, High-Performance Deep Learning Library

    Paszke, A. et al.: PyTorch: An Imperative Style, High-Performance Deep Learning Library. In: Proc. NeurIPS (2019)

    2019

  22. [22]

    In: Proc

    Kingma, D.P., Ba, J.: Adam: A Method for Stochastic Optimization. In: Proc. ICLR (2015) 18 A.T. Adib et al

    2015

  23. [23]

    et al.: Crossformer: Transformer Utilizing Cross-Dimension Dependency for Multivariate Time Series Forecasting

    Zhang, Y. et al.: Crossformer: Transformer Utilizing Cross-Dimension Dependency for Multivariate Time Series Forecasting. In: Proc. NeurIPS (2023)

    2023

  24. [24]

    et al.: TiDE: Long-term Forecasting with Time-series Dense Encoder

    Das, A. et al.: TiDE: Long-term Forecasting with Time-series Dense Encoder. In: Proc. ICML (2023)

    2023

  25. [25]

    et al.: SCINet: Time Series Modeling and Forecasting with Sample Convo- lution and Interaction

    Liu, Z. et al.: SCINet: Time Series Modeling and Forecasting with Sample Convo- lution and Interaction. In: Proc. NeurIPS (2021)

    2021

  26. [26]

    et al.: Modeling Long- and Short-Term Temporal Patterns with Deep Neu- ral Networks

    Lai, G. et al.: Modeling Long- and Short-Term Temporal Patterns with Deep Neu- ral Networks. In: Proc. SIGIR (2018)

    2018

  27. [27]

    et al.: TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods

    Qiu, X. et al.: TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods. PVLDB (2024)

    2024

  28. [28]

    et al.: Deep Time Series Models: A Comprehensive Survey and Bench- mark

    Wang, Y. et al.: Deep Time Series Models: A Comprehensive Survey and Bench- mark. arXiv:2407.13278 (2024)

    Pith/arXiv arXiv 2024

  29. [29]

    et al.: Transformers in Time Series: A Survey

    Li, C. et al.: Transformers in Time Series: A Survey. In: Proc. IJCAI (2023)

    2023

  30. [30]

    et al.: Knowledge-Empowered Dynamic Graph Network for Irregularly Sampled Medical Time Series

    Zhang, J. et al.: Knowledge-Empowered Dynamic Graph Network for Irregularly Sampled Medical Time Series. In: Proc. NeurIPS (2024)

    2024

  31. [31]

    et al.: Transformers are RNNs: Fast Autoregressive Transform- ers with Linear Attention

    Katharopoulos, A. et al.: Transformers are RNNs: Fast Autoregressive Transform- ers with Linear Attention. In: Proc. NeurIPS (2020)

    2020

  32. [32]

    Stochastic Pooling for Regularization of Deep Con- volutional Neural Networks,

    Zeiler, M. D. and Fergus, R., “Stochastic Pooling for Regularization of Deep Con- volutional Neural Networks,” inProceedings of the International Conference on Learning Representations (ICLR), 2013

    2013

  33. [33]

    CRC Press (2007)

    Madsen, H.: Time Series Analysis. CRC Press (2007)

    2007

  34. [34]

    Hendrycks, D., Gimpel, K.: Gaussian Error Linear Units (GELUs). arXiv:1606.08415 (2016) CARNet: Cycle-Conditioned Core Aggregation and Redistribution 19 A More Details on CARNet The overall architecture of CARNet is described in Algorithm 1, while the proposed Multihead core aggregation mechanism is detailed in Algorithm 2. The pseudocode formalizes the f...

    Pith/arXiv arXiv 2016