Pith. sign in

REVIEW 2 major objections 2 minor 79 references

XFMNet: Decoding Cross-Site and Nonstationary Water Patterns via Stepwise Multimodal Fusion for Long-Term Water Quality Forecasting

T0 review · 2 major / 2 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read XFMNet claims that fusing water-quality series with remote-sensing precipitation imagery, via adaptive downsampling, local decomposition, and cross-attention gated fusion, substantially improves long-term multi-site water-quality forecasts.

desk verdict The abstract describes a plausible water-quality forecasting architecture, but the submitted full text is a different paper, so the empirical claims are completely unverifiable. read the letter →

arxiv 2508.08279 v1 pith:TOIXO3ZZ submitted 2025-08-01 cs.LG cs.AI

Ziqi Wang , Hailiang Zhao , Cheng Bao , Wenzhuo Qian , Yuhao Yang , Xueqiang Sun , Shuiguang Deng This is my paper
classification cs.LGcs.AI
keywords waterqualityforecastingmultimodalfusionremotesensingprecipitationnonstationarytimeseriescross-attentiontemporaldecompositionmulti-sitepredictionlong-term
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper introduces XFMNet, a deep network for forecasting water quality at many river sites over long horizons. The central claim is that combining historical water-quality series with remote-sensing precipitation imagery, through a carefully staged fusion process, yields substantially more accurate forecasts than existing baselines. A sympathetic reader would care because water-quality monitoring increasingly relies on forecasting to guide intervention, and multi-site, nonstationary data have resisted generic time-series models. The proposed mechanism is explicit: align temporal resolutions, decompose each series into trend and cycle parts, then fuse the decomposed temporal signal with spatial precipitation context via cross-attention gated fusion.

What carries the argument

The central mechanism is a three-stage multimodal fusion pipeline. First, adaptive downsampling matches the temporal resolution of water-quality series to remote-sensing precipitation imagery. Second, locally adaptive decomposition disentangles each series into trend and cycle components, so the model can learn nonstationary behavior without assuming a fixed periodicity. Third, a cross-attention gated fusion module dynamically combines temporal patterns with spatial and ecological cues from precipitation imagery, and progressive or recursive fusion propagates both long-term trends and short-term fluctuations through the forecast horizon. This staged design is the load-bearing object: each stage is meant to remove a specific failure mode (mismatched sampling, nonstationarity, and poorly weighted modality fusion) before the next stage operates.

What would settle it

Train XFMNet on a catchment dominated by point-source pollution, where precipitation is weakly linked to water quality; if the reported improvement over a temporal-only baseline largely vanishes, the precipitation-modality premise is the load-bearing one.

Watch

Extended reading notes

Core claim

The paper's central claim is that XFMNet forecasts long-term, spatially distributed water quality more accurately than state-of-the-art baselines by learning a stepwise fusion of temporal water-quality dynamics and spatial precipitation context. Concretely, adaptive downsampling aligns the different sampling rates of in-situ water-quality readings and remote-sensing imagery; locally adaptive decomposition separates each series into trend and periodic components; and a cross-attention gated fusion module decides, per site and per time step, how much weight to give temporal patterns versus spatial and ecological cues. Progressive and recursive fusion then carries both long-term trends and short-term fluctuations into the forecast. If the experiments are sound, the result is that the model's improvement comes from jointly exploiting temporal decomposition and spatially contextual precipitation data rather than from any single component alone.

Load-bearing premise

The load-bearing premise is that remotely sensed precipitation, after temporal alignment, carries spatial and environmental information that is genuinely predictive of water quality at each river site; if that link is weak, the fusion stage has nothing useful to add.

Editorial extensions

If this is right

  • If correct, water-quality forecasting at multiple river sites can be improved by explicitly fusing remote-sensing precipitation rather than treating each site independently.
  • The decomposition-then-fusion design offers a template for other environmental forecasting tasks where observations and gridded satellite products have different temporal resolutions and nonstationary dynamics.
  • The cross-attention gated fusion should make the model robust to site-specific anomalies, since the gate can down-weight a corrupted or anomalous local series and lean on spatial context.
  • Long-horizon forecasts would carry both smooth trends and short-term fluctuations, which matters for operational systems that need warnings as well as baselines.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable extension is to ablate the precipitation imagery entirely and measure the accuracy drop; the paper's own logic predicts the drop should be largest at sites with the strongest rainfall-driven water-quality response.
  • The same architecture could be transferred to other paired data, such as air-quality monitoring with satellite aerosol optical depth, where the same temporal-alignment and cross-attention fusion logic should apply.
  • The reliance on precipitation as the spatial context suggests the method will work best in catchments where rainfall-runoff is the dominant transport pathway; in groundwater-dominated or heavily managed river reaches, the fusion gain may shrink.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The submission, arXiv:2508.08279, claims to introduce XFMNet, a stepwise multimodal fusion network for long-term, spatially distributed water quality forecasting. According to the abstract, XFMNet aligns water quality series with remote-sensing precipitation imagery via adaptive downsampling, disentangles trend and cycle components through locally adaptive decomposition, and integrates temporal, spatial, and ecological cues via cross-attention gated fusion, with the stated result of 'substantial improvements over state-of-the-art baselines.' However, the supplied full text is not an XFMNet paper at all; it is a separate manuscript titled 'Probabilistic Emissivity Retrieval from Hyperspectral Data via Physics-Guided Variational Inference' (arXiv:2508.08291v2). As a result, the reviewed material contains no architectural description, no experimental setup, no datasets, no baselines, no metrics, and no results that could substantiate the abstract's claims.

Significance. If the claimed XFMNet architecture and the associated real-world experiments are real and reproducible, the work could be a valuable contribution to water quality forecasting, particularly for multi-site settings where spatial context from remote sensing may improve predictions. The proposed combination of adaptive downsampling, locally adaptive decomposition, and cross-attention gated fusion is plausible as a technical approach. However, because the submitted full text is an entirely different paper, the significance of this manuscript cannot be assessed at all. There is no verifiable evidence in the reviewed material for either the architectural novelty or the claimed empirical gains, so the significance of the work remains entirely hypothetical to this referee.

major comments (2)
  1. [Full text] The full text of the submission is the manuscript 'Probabilistic Emissivity Retrieval from Hyperspectral Data via Physics-Guided Variational Inference' (arXiv:2508.08291v2), which has no overlap with the XFMNet abstract. None of the claimed components—adaptive downsampling, locally adaptive decomposition, cross-attention gated fusion, progressive and recursive fusion—appears in the supplied text, and there are no equations, algorithms, or architecture diagrams for XFMNet. The central claim that XFMNet improves upon state-of-the-art baselines is therefore unsupported by any element of the reviewed manuscript.
  2. [Abstract] The abstract asserts 'Extensive experiments on real-world datasets demonstrate substantial improvements over state-of-the-art baselines,' but the abstract itself does not name the datasets, baselines, evaluation metrics, number of sites, or forecast horizons, and the full text provides none of these details. Without any numerical results, error bars, or statistical tests, the improvement claim cannot be checked even in principle from the submitted material. This is a missing-support issue that affects the load-bearing empirical claim of the paper.
minor comments (2)
  1. [Abstract] The abstract introduces 'recursive fusion' as a component of the method, but the term is left undefined in the abstract and no supporting text exists to clarify it.
  2. [Abstract] The abstract describes 'adaptive downsampling' and 'locally adaptive decomposition' as key steps, but provides no explanation of how these are computed or what hyperparameters they involve; if the correct manuscript were supplied, these would need precise definitions.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation is present in the available material; the XFMNet abstract makes an empirical performance claim with no fitted-parameter-as-prediction or self-citation chain to reduce to its inputs.

full rationale

The abstract for XFMNet describes a multistage architecture (adaptive downsampling, locally adaptive decomposition, cross-attention gated fusion) and claims improved forecasting based on 'extensive experiments on real-world datasets.' No equation or derivation is given in which a fitted quantity is renamed a prediction, and no load-bearing self-citation or imported uniqueness theorem appears. The central claim is an empirical benchmark claim that is falsifiable in principle. The supplied full text is a different manuscript on probabilistic emissivity retrieval, so it cannot be used to inspect XFMNet's derivation; however, a mismatch of supplied text is a completeness/evidence limitation, not a circularity. Under the hard rule that circularity must be exhibited by quoting the paper's own reduction, no such exhibit is possible here.

Assumptions & free parameters 2 free parameters · 2 assumptions · 0 invented entities

Only the abstract was available. The architecture introduces data-dependent components but no new physical entities. The main assumptions are domain assumptions about decomposability and the informativeness of precipitation imagery.

free parameters (2)
  • Adaptive downsampling parameters = unspecified
    Used to align temporal resolutions between water quality series and remote sensing inputs; likely tuned or learned, but no values are given in the abstract.
  • Locally adaptive decomposition parameters = unspecified
    Controls trend and cycle separation; values are not provided in the abstract.
assumptions (2)
  • domain assumption Water quality time series can be decomposed into trend and cycle components.
    The abstract states that locally adaptive decomposition disentangles trend and cycle components, which presumes this decomposition is valid for the data.
  • domain assumption Remote sensing precipitation imagery provides relevant spatial and environmental context for water quality at forecast horizons.
    This is the core modeling premise behind the multimodal fusion; if false, the added modality would not improve forecasting.

how reviews work

0 comments
Cite this review

Pith. "Pith review of XFMNet: Decoding Cross-Site and Nonstationary Water Patterns via Stepwise Multimodal Fusion for Long-Term Water Quality Forecasting." pith.science (2026). https://pith.science/paper/TOIXO3ZZ

@misc{pith2026250808279,
  author       = {Pith},
  title        = {Pith review of: XFMNet: Decoding Cross-Site and Nonstationary Water Patterns via Stepwise Multimodal Fusion for Long-Term Water Quality Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TOIXO3ZZ}},
  note         = {Machine review of arXiv:2508.08279}
}
read the original abstract

Long-term time-series forecasting is critical for environmental monitoring, yet water quality prediction remains challenging due to complex periodicity, nonstationarity, and abrupt fluctuations induced by ecological factors. These challenges are further amplified in multi-site scenarios that require simultaneous modeling of temporal and spatial dynamics. To tackle this, we introduce XFMNet, a stepwise multimodal fusion network that integrates remote sensing precipitation imagery to provide spatial and environmental context in river networks. XFMNet first aligns temporal resolutions between water quality series and remote sensing inputs via adaptive downsampling, followed by locally adaptive decomposition to disentangle trend and cycle components. A cross-attention gated fusion module dynamically integrates temporal patterns with spatial and ecological cues, enhancing robustness to nonstationarity and site-specific anomalies. Through progressive and recursive fusion, XFMNet captures both long-term trends and short-term fluctuations. Extensive experiments on real-world datasets demonstrate substantial improvements over state-of-the-art baselines, highlighting the effectiveness of XFMNet for spatially distributed time series prediction.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

79 extracted references · 65 canonical work pages

  1. [1]

    Hybrid Water Quality Prediction with Multimodal Low-rank Fusion and Localized Attention , year=

    Bi, Jing and Li, Yibo and Yuan, Haitao and Wang, Mengyuan and Wang, Ziqi and Zhang, Jia and Zhou, MengChu , journal=. Hybrid Water Quality Prediction with Multimodal Low-rank Fusion and Localized Attention , year=

    work page

  2. [2]

    Long-Term Water Quality Prediction With Transformer-Based Spatial-Temporal Graph Fusion , year=

    Bi, Jing and Wang, Ziqi and Yuan, Haitao and Wu, Xiangxi and Wu, Renren and Zhang, Jia and Zhou, MengChu , journal=. Long-Term Water Quality Prediction With Transformer-Based Spatial-Temporal Graph Fusion , year=

    work page

  3. [3]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Block Hankel tensor ARIMA for multiple short time series forecasting , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  4. [4]

    ICLR 2023 Workshop on Time Series Representation Learning for Health , year=

    ForecastPFN: Universal forecasting for healthcare , author=. ICLR 2023 Workshop on Time Series Representation Learning for Health , year=

    work page 2023

  5. [5]

    A Stochastic Recurrent Encoder Decoder Network for Multistep Probabilistic Wind Power Predictions , year=

    Zheng, Zhong and Zhang, Zijun , journal=. A Stochastic Recurrent Encoder Decoder Network for Multistep Probabilistic Wind Power Predictions , year=

    work page

  6. [6]

    A Transformer-Based Industrial Time Series Prediction Model With Multivariate Dynamic Embedding , year=

    Wang, Chenze and Wang, Han and Zhang, Xiaohan and Liu, Qing and Liu, Min and Xu, Gaowei , journal=. A Transformer-Based Industrial Time Series Prediction Model With Multivariate Dynamic Embedding , year=

    work page

  7. [7]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Joint air quality and weather prediction based on multi-adversarial spatiotemporal networks , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  8. [9]

    Proceedings of the AAAI Conference on Artificial Intelligence , author=

    MSGNet: Learning Multi-Scale Inter-series Correlations for Multivariate Time Series Forecasting , volume=. Proceedings of the AAAI Conference on Artificial Intelligence , author=. 2024 , month=. doi:10.1609/aaai.v38i10.28991 , abstractNote=

    doi:10.1609/aaai.v38i10.28991 2024

Show all 79 references
  1. [10]

    The Twelfth International Conference on Learning Representations , year=

    TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting , author=. The Twelfth International Conference on Learning Representations , year=

  2. [11]

    The Twelfth International Conference on Learning Representations , year=

    iTransformer: Inverted Transformers Are Effective for Time Series Forecasting , author=. The Twelfth International Conference on Learning Representations , year=

  3. [12]

    The Eleventh International Conference on Learning Representations , year=

    TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis , author=. The Eleventh International Conference on Learning Representations , year=

  4. [13]

    Proceedings of the 39th International Conference on Machine Learning (ICML) , year=

    FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting , author=. Proceedings of the 39th International Conference on Machine Learning (ICML) , year=

  5. [15]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    FilterTS: Comprehensive Frequency Filtering for Multivariate Time Series Forecasting , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=. 2025 , doi=

    2025

  6. [16]

    Proceedings of the AAAI Conference on Artificial Intelligence , year=

    TimePFN: Effective Multivariate Time Series Forecasting with Synthetic Data , author=. Proceedings of the AAAI Conference on Artificial Intelligence , year=

  7. [17]

    Research on dynamic production plan models based on the ARMA model and Holt-Winters methods , year=

    Wang, Guixiang and Huang, Yusen and Li, Jiawen and Wei, Shuaiyang , booktitle=. Research on dynamic production plan models based on the ARMA model and Holt-Winters methods , year=

  8. [18]

    and Steve, Leander P and Narayan, Nitish and Upadhye, Abhishek and Aravindharaj, M

    Sathya, R. and Steve, Leander P and Narayan, Nitish and Upadhye, Abhishek and Aravindharaj, M. , booktitle=. Covid Wave Prediction using SARIMA Machine Learning Algorithm , year=

  9. [19]

    A process level network traffic prediction algorithm based on ARIMA model in smart substation , year=

    Wang, Jin , booktitle=. A process level network traffic prediction algorithm based on ARIMA model in smart substation , year=

  10. [20]

    Cloud Computing Resource Load Prediction Based on the Improved Particle Swarm Algorithm Optimizing GRU-RNN Model , year=

    Guo, Xiang and Hu, Yinggang and Song, Chuyi and Jiang, Jingqing and Song, Jiazhi , booktitle=. Cloud Computing Resource Load Prediction Based on the Improved Particle Swarm Algorithm Optimizing GRU-RNN Model , year=

  11. [21]

    GRU-LSTM Model Based on the SSA for Short-Term Traffic Flow Prediction , year=

    Ma, Changxi and Huang, Xiaoyu and Zhao, Yongpeng and Wang, Tao and Du, Bo , journal=. GRU-LSTM Model Based on the SSA for Short-Term Traffic Flow Prediction , year=

  12. [22]

    STAM: A SpatioTemporal Attention Based Memory for Video Prediction , year=

    Chang, Zheng and Zhang, Xinfeng and Wang, Shanshe and Ma, Siwei and Gao, Wen , journal=. STAM: A SpatioTemporal Attention Based Memory for Video Prediction , year=

  13. [23]

    Short-Term Wind Power Prediction Based on STL-AOA-Transformer Algorithm , year=

    Li, Zhifeng and Ye, Yurun and Liu, Wenxin and Lu, Ang , booktitle=. Short-Term Wind Power Prediction Based on STL-AOA-Transformer Algorithm , year=

  14. [24]

    One-Sided Relational Autoencoder With Seasonal-Trend Decomposition to Extract Process Correlations for Molten Iron Quality Prediction , year=

    Yan, Duojin and Yang, Chunjie and Sun, Shuming and Lou, Siwei and Kong, Liyuan and Zhang, Yuyan , journal=. One-Sided Relational Autoencoder With Seasonal-Trend Decomposition to Extract Process Correlations for Molten Iron Quality Prediction , year=

  15. [25]

    Modeling Method for Wind Farm Based on Equivalent Dynamic Response , year=

    Zhang, Haifeng and Zhang, Yifu and Jiang, Dongxiao and Leng, Jun and Zhang, Zongbao and Peng, Xiaoyu , booktitle=. Modeling Method for Wind Farm Based on Equivalent Dynamic Response , year=

  16. [26]

    Feature-Level Fusion vs

    Neshov, Nikolay and Tonchev, Krasmir and Manolova, Agata and Poulkov, Vladimir and Balabanov, Georgi , journal=. Feature-Level Fusion vs. Score-Level Fusion for Image Retrieval Based on Pre-Trained Deep Neural Networks , year=

  17. [27]

    ID-DA Fusion:Infrared and Visible Image Fusion Based on Intermediate Domain and Dual Attention Mechanism , year=

    Liu, Shuo and Zhou, Huabing , booktitle=. ID-DA Fusion:Infrared and Visible Image Fusion Based on Intermediate Domain and Dual Attention Mechanism , year=

  18. [28]

    U-Shape Spatial-Temporal Prediction Network Based on 3D Convolution and BDLSTM , year=

    Peng, Ge and Shi, Chunchao and Zhong, Yujing and Ai, Xinyu , booktitle=. U-Shape Spatial-Temporal Prediction Network Based on 3D Convolution and BDLSTM , year=

  19. [29]

    Predict-and-Optimize Model for Day-Ahead Inertia Prediction Using Distributionally Robust Unit Commitment With Renewable Energy Sources , year=

    Shen, Fu and Cao, Yang and Shahidehpour, Mohammad and Xu, Xiaoyuan and Wang, Chong and Wang, Jian and Zhai, Suwei , journal=. Predict-and-Optimize Model for Day-Ahead Inertia Prediction Using Distributionally Robust Unit Commitment With Renewable Energy Sources , year=

  20. [30]

    Identifying the origin of the magnetic directional anomalies recorded in the Datong loess profile, northeastern Chinese loess plateau , year=

    Zhu, Rixiang and Liu, Qingsong and Pan, Yongxin and Deng, Chenglong and Sun, Jimin , journal=. Identifying the origin of the magnetic directional anomalies recorded in the Datong loess profile, northeastern Chinese loess plateau , year=

  21. [31]

    International conference on machine learning , pages=

    Efficientnet: Rethinking model scaling for convolutional neural networks , author=. International conference on machine learning , pages=. 2019 , organization=

    2019

  22. [32]

    Deep CNN Feature Resampling and Ensemble Based on Cross Validation for Image Classification , year=

    Wang, Yu and Zhang, Haodong and Yang, Xingli and Li, Jihong , journal=. Deep CNN Feature Resampling and Ensemble Based on Cross Validation for Image Classification , year=

  23. [33]

    Beam Prediction Based on Large Language Models , year=

    Sheng, Yucheng and Huang, Kai and Liang, Le and Liu, Peng and Jin, Shi and Ye Li, Geoffrey , journal=. Beam Prediction Based on Large Language Models , year=

  24. [34]

    and Zheng, X

    Zhou, N. and Zheng, X. and He, D. and Hong, D. and Chanussot, J. , title =. IEEE Transactions on Pattern Analysis and Machine Intelligence , year =. doi:10.1109/TPAMI.2024.3475485 , keywords =

    2024

  25. [35]

    and Schwartz, Mischa , booktitle=

    Richman, Steven H. and Schwartz, Mischa , booktitle=. Dynamic programming training period for an MSE adaptive equalizer , year=

  26. [36]

    and Boeri, F

    Menez, J. and Boeri, F. and Esteban, D. , booktitle=. Optimum quantizer algorithm for real-time block quantizing , year=

  27. [37]

    Attention Bottlenecks for Multimodal Fusion , url =

    Nagrani, Arsha and Yang, Shan and Arnab, Anurag and Jansen, Aren and Schmid, Cordelia and Sun, Chen , booktitle =. Attention Bottlenecks for Multimodal Fusion , url =

  28. [38]

    International Journal of Computer Vision , volume=

    Mutualformer: Multi-modal representation learning via cross-diffusion attention , author=. International Journal of Computer Vision , volume=. 2024 , publisher=

    2024

  29. [39]

    Sleep Stage Classification using Multimodal Embedding Fusion from Electrooculography and Pressure-Sensitive Mats , year=

    Papillon, Olivier and Goubran, Rafik and Green, James and Larivière-Chartier, Julien and Higginson, Caitlin and Knoefel, Frank and Robillard, Rébecca , booktitle=. Sleep Stage Classification using Multimodal Embedding Fusion from Electrooculography and Pressure-Sensitive Mats , year=

  30. [40]

    LMF-Net: A Learnable Multimodal Fusion Network for Semantic Segmentation of Remote Sensing Data , year=

    Li, Jihao and Zhang, Wenkai and Zhang, Weihang and Zhou, Ruixue and Li, Chongyang and Tong, Boyuan and Sun, Xian and Fu, Kun , journal=. LMF-Net: A Learnable Multimodal Fusion Network for Semantic Segmentation of Remote Sensing Data , year=

  31. [41]

    MSA-TFN: Multi-Scale Attention Two-step Fusion Network for EEG-fNIRS Motor Imagery Classification , year=

    Kang, Ranlan and Li, Yurong , booktitle=. MSA-TFN: Multi-Scale Attention Two-step Fusion Network for EEG-fNIRS Motor Imagery Classification , year=

  32. [43]

    CoRR , year=

    Timer: Transformers for time series analysis at scale , author=. CoRR , year=

  33. [44]

    Advances in Neural Information Processing Systems , volume=

    Autotimes: Autoregressive time series forecasters via large language models , author=. Advances in Neural Information Processing Systems , volume=

  34. [45]

    arXiv preprint arXiv:2402.04852 , year=

    Multi-Patch Prediction: Adapting LLMs for Time Series Representation Learning , author=. arXiv preprint arXiv:2402.04852 , year=. doi:10.48550/arXiv.2402.04852 , url=. 2402.04852 , archivePrefix=

    doi arXiv

  35. [46]

    Bi, J.; Wang, Z.; Yuan, H.; Wu, X.; Wu, R.; Zhang, J.; and Zhou, M. 2025. Long-Term Water Quality Prediction With Transformer-Based Spatial-Temporal Graph Fusion. IEEE Transactions on Automation Science and Engineering, 22: 11392--11404

    2025

  36. [47]

    Bian, Y.; Ju, X.; Li, J.; Xu, Z.; Cheng, D.; and Xu, Q. 2024. Multi-Patch Prediction: Adapting LLMs for Time Series Representation Learning. arXiv preprint arXiv:2402.04852. Version 2 (Updated March 10, 2024)

    2024 arXiv

  37. [48]

    Cai, W.; Liang, Y.; Liu, X.; Feng, J.; and Wu, Y. 2024. MSGNet: Learning Multi-Scale Inter-series Correlations for Multivariate Time Series Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence, 38(10): 11141--11149

    2024

  38. [49]

    Guo, X.; Hu, Y.; Song, C.; Jiang, J.; and Song, J. 2024. Cloud Computing Resource Load Prediction Based on the Improved Particle Swarm Algorithm Optimizing GRU-RNN Model. In 2024 20th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-...

    2024

  39. [50]

    Han, J.; Liu, H.; Zhu, H.; Xiong, H.; and Dou, D. 2021. Joint air quality and weather prediction based on multi-adversarial spatiotemporal networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, 4081--4089

    2021

  40. [51]

    Huang, S.; Zhao, Z.; Li, C.; and Bai, L. 2025. TimeKAN: KAN-based Frequency Decomposition Learning Architecture for Long-term Time Series Forecasting. arXiv preprint arXiv:2502.06910

    2025 arXiv

  41. [52]

    Kang, R.; and Li, Y. 2024. MSA-TFN: Multi-Scale Attention Two-step Fusion Network for EEG-fNIRS Motor Imagery Classification. In 2024 2nd International Conference on Computer Network Technology and Electronic and Information Engineering (CNTEIE), 64--68

    2024

  42. [53]

    Kim, K.; Tsai, H.; Sen, R.; Das, A.; Zhou, Z.; Tanpure, A.; Luo, M.; and Yu, R. 2024. Multi-Modal Forecaster: Jointly Predicting Time Series and Textual Data. arXiv preprint arXiv:2411.06735

    2024 arXiv

  43. [54]

    Li, J.; Zhang, W.; Zhang, W.; Zhou, R.; Li, C.; Tong, B.; Sun, X.; and Fu, K. 2025. LMF-Net: A Learnable Multimodal Fusion Network for Semantic Segmentation of Remote Sensing Data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 18: 3905--3920

    2025

  44. [55]

    Li, Z.; Ye, Y.; Liu, W.; and Lu, A. 2024. Short-Term Wind Power Prediction Based on STL-AOA-Transformer Algorithm. In 2024 The 9th International Conference on Power and Renewable Energy (ICPRE), 1390--1395

    2024

  45. [56]

    Liu, Y.; Hu, T.; Zhang, H.; Wu, H.; Wang, S.; Ma, L.; and Long, M. 2024 a . iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. In The Twelfth International Conference on Learning Representations

    2024

  46. [57]

    Liu, Y.; Qin, G.; Huang, X.; Wang, J.; and Long, M. 2024 b . Autotimes: Autoregressive time series forecasters via large language models. Advances in Neural Information Processing Systems, 37: 122154--122184

    2024

  47. [58]

    Liu, Y.; Zhang, H.; Li, C.; Huang, X.; Wang, J.; and Long, M. 2024 c . Timer: Transformers for time series analysis at scale. CoRR

    2024

  48. [59]

    Ma, C.; Huang, X.; Zhao, Y.; Wang, T.; and Du, B. 2025. GRU-LSTM Model Based on the SSA for Short-Term Traffic Flow Prediction. Journal of Intelligent and Connected Vehicles, 8(1): 9210051--1--9210051--10

    2025

  49. [60]

    Neshov, N.; Tonchev, K.; Manolova, A.; Poulkov, V.; and Balabanov, G. 2024. Feature-Level Fusion vs. Score-Level Fusion for Image Retrieval Based on Pre-Trained Deep Neural Networks. Journal of Mobile Multimedia, 20(4): 769--783

    2024

  50. [61]

    Papillon, O.; Goubran, R.; Green, J.; Larivière-Chartier, J.; Higginson, C.; Knoefel, F.; and Robillard, R. 2025. Sleep Stage Classification using Multimodal Embedding Fusion from Electrooculography and Pressure-Sensitive Mats. In 2025 IEEE Medical Measurements and Application...

    2025

  51. [62]

    Peng, G.; Shi, C.; Zhong, Y.; and Ai, X. 2024. U-Shape Spatial-Temporal Prediction Network Based on 3D Convolution and BDLSTM. In 2024 IEEE 4th International Conference on Software Engineering and Artificial Intelligence (SEAI), 257--261

    2024

  52. [63]

    P.; Narayan, N.; Upadhye, A.; and Aravindharaj, M

    Sathya, R.; Steve, L. P.; Narayan, N.; Upadhye, A.; and Aravindharaj, M. 2023. Covid Wave Prediction using SARIMA Machine Learning Algorithm. In 2023 IEEE International Students' Conference on Electrical, Electronics and Computer Science (SCEECS), 1--5

    2023

  53. [64]

    Shen, F.; Cao, Y.; Shahidehpour, M.; Xu, X.; Wang, C.; Wang, J.; and Zhai, S. 2025. Predict-and-Optimize Model for Day-Ahead Inertia Prediction Using Distributionally Robust Unit Commitment With Renewable Energy Sources. IEEE Transactions on Power Systems, 40(3): 2688--2699

    2025

  54. [65]

    Sheng, Y.; Huang, K.; Liang, L.; Liu, P.; Jin, S.; and Ye Li, G. 2025. Beam Prediction Based on Large Language Models. IEEE Wireless Communications Letters, 14(5): 1406--1410

    2025

  55. [66]

    Shi, Q.; Yin, J.; Cai, J.; Cichocki, A.; Yokota, T.; Chen, L.; Yuan, M.; and Zeng, J. 2020. Block Hankel tensor ARIMA for multiple short time series forecasting. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, 5758--5766

    2020

  56. [67]

    O.; Ildiz, M

    Taga, E. O.; Ildiz, M. E.; and Oymak, S. 2025. TimePFN: Effective Multivariate Time Series Forecasting with Synthetic Data. In Proceedings of the AAAI Conference on Artificial Intelligence. To appear

    2025

  57. [68]

    Wang, C.; Wang, H.; Zhang, X.; Liu, Q.; Liu, M.; and Xu, G. 2025 a . A Transformer-Based Industrial Time Series Prediction Model With Multivariate Dynamic Embedding. IEEE Transactions on Industrial Informatics, 21(2): 1813--1822

    2025

  58. [69]

    Wang, G.; Huang, Y.; Li, J.; and Wei, S. 2023. Research on dynamic production plan models based on the ARMA model and Holt-Winters methods. In 2023 8th International Conference on Information Systems Engineering (ICISE), 221--225

    2023

  59. [70]

    Wang, J. 2013. A process level network traffic prediction algorithm based on ARIMA model in smart substation. In 2013 IEEE International Conference on Signal Processing, Communication and Computing (ICSPCC 2013), 1--5

    2013

  60. [71]

    Y.; and ZHOU, J

    Wang, S.; Wu, H.; Shi, X.; Hu, T.; Luo, H.; Ma, L.; Zhang, J. Y.; and ZHOU, J. 2024 a . TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting. In The Twelfth International Conference on Learning Representations

    2024

  61. [72]

    Wang, X.; Wang, X.; Jiang, B.; Tang, J.; and Luo, B. 2024 b . Mutualformer: Multi-modal representation learning via cross-diffusion attention. International Journal of Computer Vision, 132(9): 3867--3888

    2024

  62. [73]

    Wang, Y.; Liu, Y.; Duan, X.; and Wang, K. 2025 b . FilterTS: Comprehensive Frequency Filtering for Multivariate Time Series Forecasting. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, 35438--35445

    2025

  63. [74]

    Wang, Y.; Zhang, H.; Yang, X.; and Li, J. 2025 c . Deep CNN Feature Resampling and Ensemble Based on Cross Validation for Image Classification. IEEE Transactions on Neural Networks and Learning Systems, 36(6): 10899--10912

    2025

  64. [75]

    Wu, H.; Hu, T.; Liu, Y.; Zhou, H.; Wang, J.; and Long, M. 2023. TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis. In The Eleventh International Conference on Learning Representations

    2023

  65. [76]

    Yan, D.; Yang, C.; Sun, S.; Lou, S.; Kong, L.; and Zhang, Y. 2024. One-Sided Relational Autoencoder With Seasonal-Trend Decomposition to Extract Process Correlations for Molten Iron Quality Prediction. IEEE Transactions on Instrumentation and Measurement, 73: 1--13

    2024

  66. [77]

    Zhang, H.; Zhang, Y.; Jiang, D.; Leng, J.; Zhang, Z.; and Peng, X. 2023. Modeling Method for Wind Farm Based on Equivalent Dynamic Response. In 2023 3rd International Conference on Energy Engineering and Power Systems (EEPS), 377--381

    2023

  67. [78]

    Zheng, Z.; and Zhang, Z. 2024. A Stochastic Recurrent Encoder Decoder Network for Multistep Probabilistic Wind Power Predictions. IEEE Transactions on Neural Networks and Learning Systems, 35(7): 9565--9578

    2024

  68. [79]

    Zhong, S.; Ruan, W.; Jin, M.; Li, H.; Wen, Q.; and Liang, Y. 2025. Time-vlm: Exploring multimodal vision-language models for augmented time series forecasting. arXiv preprint arXiv:2502.04395

    2025 arXiv

  69. [80]

    Zhou, N.; Zheng, X.; He, D.; Hong, D.; and Chanussot, J. 2025. Probing Synergistic High-Order Interaction for Multi-Modal Image Fusion. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(2): 840--857

    2025

  70. [81]

    Zhou, T.; Ma, Z.; Wen, Q.; Wang, X.; Sun, L.; and Jin, R. 2022. FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting. In Proceedings of the 39th International Conference on Machine Learning (ICML). ArXiv:2201.12740

    2022 arXiv

  71. [82]

    Zhu, R.; Liu, Q.; Pan, Y.; Deng, C.; and Sun, J. 2006. Identifying the origin of the magnetic directional anomalies recorded in the Datong loess profile, northeastern Chinese loess plateau. Geophysical Journal International, 164(2): 312--318

    2006

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.