pith. sign in

arxiv: 2509.23668 · v3 · submitted 2025-09-28 · 💻 cs.LG

Hermes: A Multi-Scale Spatial-Temporal Hypergraph Network for Stock Time Series Forecasting

Xiangfei Qiu , Liu Yang , Xiangyu Xu , Hanyin Cheng , Xingjian Wu , Rongjia Wu , Zhigang Zhang , Ding Tu
show 4 more authors
Chenjuan Guo Bin Yang Christian S. Jensen Jilin Hu
This is my paper

Pith reviewed 2026-05-18 12:30 UTC · model grok-4.3

classification 💻 cs.LG
keywords stock time series forecastinghypergraph neural networkslead-lag interactionsmulti-scale modelingindustry correlationstemporal aggregation
0
0 comments X p. Extension

The pith

Hermes improves stock time series forecasts by capturing inter-industry lead-lag effects and multi-scale patterns through a hypergraph network.

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

Stock time series are shaped by correlations across industries, including cases where movements in one sector lead or lag others and where patterns appear at different time scales. Prior hypergraph methods capture these relationships only superficially and miss both the dynamic lead-lag structure and the multi-scale organization. Hermes introduces a hyperedge-based moving aggregation module that applies sliding windows and dynamic temporal aggregation to model flexible lead-lag dependencies among industries. It pairs this with cross-scale edge-to-edge message passing that fuses information across scales while keeping each scale internally consistent. Experiments on real stock datasets show the resulting model outperforms existing state-of-the-art forecasting methods.

Core claim

The Hermes framework improves exploitation of industry correlation in stock time series by integrating a hyperedge-based moving aggregation module, which uses a sliding window and dynamic temporal aggregation to capture inter-industry lead-lag interactions, together with cross-scale edge-to-edge message passing that integrates multi-scale information while preserving scale consistency.

What carries the argument

Hyperedge-based moving aggregation with sliding windows for lead-lag capture, combined with cross-scale edge-to-edge message passing for multi-scale fusion inside a spatial-temporal hypergraph network.

If this is right

  • More accurate modeling of dynamic temporal dependencies across industries in multivariate financial data.
  • Consistent integration of information at different temporal scales without loss of scale-specific structure.
  • Improved forecasting performance on real stock datasets compared with prior hypergraph and graph-based baselines.
  • Better support for decision-making by investors and analysts who rely on industry-level correlation signals.

Where Pith is reading between the lines

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

  • The same lead-lag and multi-scale mechanisms could be tested on other hierarchical time series such as supply-chain or commodity price data.
  • Combining the hypergraph structure with attention-based temporal encoders might further refine the capture of long-range dependencies.
  • Datasets that explicitly annotate industry hierarchies would allow direct measurement of how much the hypergraph representation contributes versus the aggregation rules.

Load-bearing premise

The accuracy gains come primarily from the two new modules rather than from other modeling choices or from particular properties of the stock datasets.

What would settle it

An ablation study that removes the moving aggregation module or the cross-scale message passing component and measures the resulting drop in forecasting accuracy on the same real-world stock datasets.

Figures

Figures reproduced from arXiv: 2509.23668 by Bin Yang, Chenjuan Guo, Christian S. Jensen, Ding Tu, Hanyin Cheng, Jilin Hu, Liu Yang, Rongjia Wu, Xiangfei Qiu, Xiangyu Xu, Xingjian Wu, Zhigang Zhang.

Figure 1
Figure 1. Figure 1: (a): Intra-industry correlation. Relations between stocks within an industry (hyperedge) [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The architecture of Hermes. stock time series forecasting model that specifically takes into account the multi-scale nature of stock data, enabling it to capture the complexity of market dynamics more comprehensively. Furthermore, our approach places particular emphasis on the lead-lag relationships between financial industries, where the changes in one industry may precede or lag behind those in another. … view at source ↗
Figure 4
Figure 4. Figure 4: Three types of message passing: hy￾pernodes for stocks and hyperedges for indus￾tries. 8 16 32 64 Lookback window length 0.025 0.030 0.035 0.040 0.045 0.050 Information Coefficient (IC) NYSDAQ NYSE S&P500 (a) Lookback length 5 8 32 64 Latent space dimension 0.020 0.025 0.030 0.035 0.040 0.045 0.050 Information Coefficient (IC) NYSDAQ NYSE S&P500 (b) Latent space dimension [3, 3, 3] [9, 6, 3] [12, 4, 3] [12… view at source ↗
Figure 5
Figure 5. Figure 5: Parameter sensitivity studies of main hyper-parameters in Hermes. [PITH_FULL_IMAGE:figures/full_fig_p020_5.png] view at source ↗
read the original abstract

Time series forecasting occurs in a range of financial applications providing essential decision-making support to investors, regulatory institutions, and analysts. Unlike multivariate time series from other domains, stock time series exhibit industry correlation. Exploiting this kind of correlation can improve forecasting accuracy. However, existing methods based on hypergraphs can only capture industry correlation relatively superficially. These methods face two key limitations: they do not fully consider inter-industry lead-lag interactions, and they do not model multi-scale information within and among industries. This study proposes the Hermes framework for stock time series forecasting that aims to improve the exploitation of industry correlation by addressing these limitations. The framework integrates moving aggregation and multi-scale fusion modules in a hypergraph network. Specifically, to more flexibly capture the lead-lag relationships among industries, Hermes proposes a hyperedge-based moving aggregation module. This module incorporates a sliding window and utilizes dynamic temporal aggregation operations to consider lead-lag dependencies among industries. Additionally, to effectively model multi-scale information, Hermes employs cross-scale, edge-to-edge message passing to integrate information from different scales while maintaining the consistency of each scale. Experimental results on multiple real-world stock datasets show that Hermes outperforms existing state-of-the-art methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes Hermes, a multi-scale spatial-temporal hypergraph network for stock time series forecasting. It identifies limitations in prior hypergraph methods for capturing industry correlations—specifically, insufficient modeling of inter-industry lead-lag interactions and inadequate handling of multi-scale information within and across industries. The framework introduces a hyperedge-based moving aggregation module that uses a sliding window and dynamic temporal aggregation to address lead-lag dependencies, combined with cross-scale edge-to-edge message passing to integrate multi-scale information while preserving scale consistency. The central empirical claim is that these components enable Hermes to outperform existing state-of-the-art methods on multiple real-world stock datasets.

Significance. If the reported performance gains prove robust and are shown to arise specifically from the proposed modules rather than ancillary design choices, the work would advance hypergraph-based modeling for financial time series by providing a more flexible treatment of temporal lead-lag effects and hierarchical multi-scale structures. This could have practical value in a domain where even modest accuracy improvements carry economic significance.

major comments (2)
  1. [Experimental Results] Experimental section: The manuscript claims that Hermes outperforms SOTA methods on real-world stock datasets, yet no ablation results are presented that remove the hyperedge-based moving aggregation module (while holding hypergraph construction, loss, optimizer, and other components fixed) to isolate its contribution to any observed gains. Without such controls, attribution of improvements to the lead-lag modeling remains unverified and load-bearing for the central claim.
  2. [Methodology] Methodology section on multi-scale fusion: The cross-scale edge-to-edge message passing is presented as integrating information across scales while maintaining consistency, but the manuscript does not detail how the distinct scales are constructed or partitioned in the hypergraph (e.g., criteria for within-industry vs. among-industry scales or the number of scales). This definition is necessary to evaluate whether the module genuinely addresses the stated multi-scale limitation.
minor comments (2)
  1. [Abstract] The abstract summarizes the proposed modules and claims but omits any quantitative performance metrics or dataset names, which would aid immediate assessment even if full tables appear later.
  2. [Notation and Equations] Notation for hyperedges, sliding-window parameters, and scale indices should be introduced once and used consistently to prevent ambiguity in the equations describing aggregation and message passing.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback on our manuscript. We address each major comment below and outline the revisions we will make to strengthen the paper.

read point-by-point responses

  1. Referee: Experimental section: The manuscript claims that Hermes outperforms SOTA methods on real-world stock datasets, yet no ablation results are presented that remove the hyperedge-based moving aggregation module (while holding hypergraph construction, loss, optimizer, and other components fixed) to isolate its contribution to any observed gains. Without such controls, attribution of improvements to the lead-lag modeling remains unverified and load-bearing for the central claim.

    Authors: We agree that an ablation isolating the hyperedge-based moving aggregation module, with all other components held fixed, is necessary to rigorously attribute performance gains to the lead-lag modeling. We will add this controlled ablation experiment to the experimental section of the revised manuscript and report the corresponding results. revision: yes

  2. Referee: Methodology section on multi-scale fusion: The cross-scale edge-to-edge message passing is presented as integrating information across scales while maintaining consistency, but the manuscript does not detail how the distinct scales are constructed or partitioned in the hypergraph (e.g., criteria for within-industry vs. among-industry scales or the number of scales). This definition is necessary to evaluate whether the module genuinely addresses the stated multi-scale limitation.

    Authors: We acknowledge that explicit details on scale construction and partitioning are required for full reproducibility and evaluation. We will expand the methodology section to describe the criteria for defining within-industry versus among-industry scales, the number of scales used, and the partitioning approach in the hypergraph. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical architecture proposal with no derivational reduction

full rationale

The paper proposes the Hermes hypergraph framework with two new modules (hyperedge-based moving aggregation using sliding windows for lead-lag, and cross-scale edge-to-edge message passing) to better exploit industry correlations in stock time series. The central claim rests on experimental outperformance versus SOTA on real-world datasets. No equations, closed-form predictions, or first-principles derivations are presented that could reduce to fitted parameters or self-citations by construction. The work is self-contained as an empirical modeling contribution; any attribution of gains to the specific modules would require ablations (not a circularity concern). No load-bearing self-citation chains or ansatz smuggling appear in the provided text.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The paper introduces a new neural architecture whose performance claims rest on standard deep-learning assumptions plus two domain-specific modeling choices whose justification is not visible in the abstract.

free parameters (2)
  • hypergraph construction parameters
    Industry correlation thresholds or hyperedge definitions are not specified and must be chosen or learned.
  • sliding window size and aggregation weights
    These control the lead-lag modeling and are fitted or tuned on data.
axioms (1)
  • domain assumption Industry sectors form meaningful hyperedges that reflect lead-lag relationships in stock returns.
    Invoked when the authors state that exploiting industry correlation improves forecasting.

pith-pipeline@v0.9.0 · 5781 in / 1289 out tokens · 38098 ms · 2026-05-18T12:30:00.526645+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

82 extracted references · 82 canonical work pages · 3 internal anchors

  1. [1]

    Robust deep learning model with attention framework for spatiotemporal forecasting of solar and wind energy production

    Md Shadman Abid, Razzaqul Ahshan, Mohammed Al-Abri, and Rashid Al Abri. Robust deep learning model with attention framework for spatiotemporal forecasting of solar and wind energy production. Energy Conversion and Management: X, pp.\ 100919, 2025

    work page 2025

  2. [2]

    Deep learning for stock prediction using numerical and textual information

    Ryo Akita, Akira Yoshihara, Takashi Matsubara, and Kuniaki Uehara. Deep learning for stock prediction using numerical and textual information. In ICIS, pp.\ 1--6, 2016

    work page 2016

  3. [3]

    Stock price prediction using k-nearest neighbor (knn) algorithm

    Khalid Alkhatib, Hassan Najadat, Ismail Hmeidi, and Mohammed K Ali Shatnawi. Stock price prediction using k-nearest neighbor (knn) algorithm. IJBHT, 3 0 (3): 0 32--44, 2013

    work page 2013

  4. [4]

    Adebiyi Ariyo Ariyo, Aderemi Oluyinka Adewumi, and Charles K. Ayo. Stock price prediction using the ARIMA model. In ICCMS, pp.\ 106--112, 2014

    work page 2014

  5. [5]

    Hypergraph attention networks

    Chaofan Chen, Zelei Cheng, Zuotian Li, and Manyi Wang. Hypergraph attention networks. In TrustCom, pp.\ 1560--1565, 2020

    work page 2020

  6. [6]

    Periodicity decoupling framework for long-term series forecasting

    Tao Dai, Beiliang Wu, Peiyuan Liu, Naiqi Li, Jigang Bao, Yong Jiang, and Shu-Tao Xia. Periodicity decoupling framework for long-term series forecasting. In ICLR, 2024 a

    work page 2024

  7. [7]

    Ddn: Dual-domain dynamic normalization for non-stationary time series forecasting

    Tao Dai, Beiliang Wu, Peiyuan Liu, Naiqi Li, Xue Yuerong, Shu-Tao Xia, and Zexuan Zhu. Ddn: Dual-domain dynamic normalization for non-stationary time series forecasting. In NeurIPS, volume 37, pp.\ 108490--108517, 2024 b

    work page 2024

  8. [8]

    On the network topology of variance decompositions: Measuring the connectedness of financial firms

    Francis X Diebold and Kamil Y lmaz. On the network topology of variance decompositions: Measuring the connectedness of financial firms. Journal of econometrics, 182 0 (1): 0 119--134, 2014

    work page 2014

  9. [9]

    Modern portfolio theory and investment analysis

    Edwin J Elton, Martin J Gruber, Stephen J Brown, and William N Goetzmann. Modern portfolio theory and investment analysis. 2009

    work page 2009

  10. [10]

    Forecasting big time series: old and new

    Christos Faloutsos, Jan Gasthaus, Tim Januschowski, and Yuyang Wang. Forecasting big time series: old and new. In Proc. VLDB Endow. , pp.\ 2102--2105, 2018

    work page 2018

  11. [11]

    Stockmixer: A simple yet strong mlp-based architecture for stock price forecasting

    Jinyong Fan and Yanyan Shen. Stockmixer: A simple yet strong mlp-based architecture for stock price forecasting. In AAAI, pp.\ 8389--8397, 2024

    work page 2024

  12. [12]

    Enhancing stock movement prediction with adversarial training

    Fuli Feng, Huimin Chen, Xiangnan He, Ji Ding, Maosong Sun, and Tat - Seng Chua. Enhancing stock movement prediction with adversarial training. In IJCAI, pp.\ 5843--5849, 2019 a

    work page 2019

  13. [13]

    Temporal relational ranking for stock prediction

    Fuli Feng, Xiangnan He, Xiang Wang, Cheng Luo, Yiqun Liu, and Tat-Seng Chua. Temporal relational ranking for stock prediction. TOIS, 37 0 (2): 0 1--30, 2019 b

    work page 2019

  14. [14]

    Hypergraph neural networks

    Yifan Feng, Haoxuan You, Zizhao Zhang, Rongrong Ji, and Yue Gao. Hypergraph neural networks. In AAAI, pp.\ 3558--3565, 2019 c

    work page 2019

  15. [15]

    Pair: Complementarity-guided disentanglement for composed image retrieval

    Zhiheng Fu, Zixu Li, Zhiwei Chen, Chunxiao Wang, Xuemeng Song, Yupeng Hu, and Liqiang Nie. Pair: Complementarity-guided disentanglement for composed image retrieval. In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing, pp.\ 1--5. IEEE, 2025

    work page 2025

  16. [16]

    Stock market prediction using hidden markov models

    Aditya Gupta and Bhuwan Dhingra. Stock market prediction using hidden markov models. In SCES, pp.\ 1--4, 2012

    work page 2012

  17. [17]

    Long short-term memory

    Sepp Hochreiter and J \"u rgen Schmidhuber. Long short-term memory. Neural computation, 9 0 (8): 0 1735--1780, 1997

    work page 1997

  18. [18]

    Adaptive multi-scale decomposition framework for time series forecasting

    Yifan Hu, Peiyuan Liu, Peng Zhu, Dawei Cheng, and Tao Dai. Adaptive multi-scale decomposition framework for time series forecasting. In AAAI, volume 39, pp.\ 17359--17367, 2025

    work page 2025

  19. [19]

    Median: Adaptive intermediate-grained aggregation network for composed image retrieval

    Qinlei Huang, Zhiwei Chen, Zixu Li, Chunxiao Wang, Xuemeng Song, Yupeng Hu, and Liqiang Nie. Median: Adaptive intermediate-grained aggregation network for composed image retrieval. In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing, pp.\ 1--5. IEEE, 2025

    work page 2025

  20. [20]

    Efficient integration of multi-order dynamics and internal dynamics in stock movement prediction

    Thanh Trung Huynh, Minh Hieu Nguyen, Thanh Tam Nguyen, Phi Le Nguyen, Matthias Weidlich, Quoc Viet Hung Nguyen, and Karl Aberer. Efficient integration of multi-order dynamics and internal dynamics in stock movement prediction. In WSDM, pp.\ 850--858, 2023

    work page 2023

  21. [21]

    Fintime: a financial time series benchmark

    Kaippallimalil J Jacob and Dennis Shasha. Fintime: a financial time series benchmark. ACM SIGMOD Record, 28 0 (4): 0 42--48, 1999

    work page 1999

  22. [22]

    Financial time series forecasting using support vector machines

    Kyoung-jae Kim. Financial time series forecasting using support vector machines. Neurocomputing, 55 0 (1-2): 0 307--319, 2003

    work page 2003

  23. [23]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In ICLR, 2015

    work page 2015

  24. [24]

    Back to basics: The power of the multilayer perceptron in financial time series forecasting

    Ana Lazcano, Miguel A Jaramillo-Mor \'a n, and Julio E Sandubete. Back to basics: The power of the multilayer perceptron in financial time series forecasting. Mathematics, 12 0 (12): 0 1920, 2024

    work page 1920

  25. [25]

    Set you straight: Auto-steering denoising trajectories to sidestep unwanted concepts

    Leyang Li, Shilin Lu, Yan Ren, and Adams Wai-Kin Kong. Set you straight: Auto-steering denoising trajectories to sidestep unwanted concepts. arXiv preprint arXiv:2504.12782, 2025 a

    work page arXiv 2025

  26. [26]

    Master: Market-guided stock transformer for stock price forecasting

    Tong Li, Zhaoyang Liu, Yanyan Shen, Xue Wang, Haokun Chen, and Sen Huang. Master: Market-guided stock transformer for stock price forecasting. In AAAI, pp.\ 162--170, 2024

    work page 2024

  27. [27]

    Modeling the stock relation with graph network for overnight stock movement prediction

    Wei Li, Ruihan Bao, Keiko Harimoto, Deli Chen, Jingjing Xu, and Qi Su. Modeling the stock relation with graph network for overnight stock movement prediction. In IJCAI, pp.\ 4541--4547, 2021

    work page 2021

  28. [28]

    Multi-modal large language model with rag strategies in soccer commentary generation

    Xiang Li, Yangfan He, Shuaishuai Zu, Zhengyang Li, Tianyu Shi, Yiting Xie, and Kevin Zhang. Multi-modal large language model with rag strategies in soccer commentary generation. In WACV, pp.\ 6197--6206, 2025 b

    work page 2025

  29. [29]

    Hypergraph-based reinforcement learning for stock portfolio selection

    Xiaojie Li, Chaoran Cui, Donglin Cao, Juan Du, and Chunyun Zhang. Hypergraph-based reinforcement learning for stock portfolio selection. In ICASSP, pp.\ 4028--4032, 2022

    work page 2022

  30. [30]

    Encoder: Entity mining and modification relation binding for composed image retrieval

    Zixu Li, Zhiwei Chen, Haokun Wen, Zhiheng Fu, Yupeng Hu, and Weili Guan. Encoder: Entity mining and modification relation binding for composed image retrieval. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pp.\ 5101--5109, 2025 c

    work page 2025

  31. [31]

    Finecir: Explicit parsing of fine-grained modification semantics for composed image retrieval

    Zixu Li, Zhiheng Fu, Yupeng Hu, Zhiwei Chen, Haokun Wen, and Liqiang Nie. Finecir: Explicit parsing of fine-grained modification semantics for composed image retrieval. https://arxiv.org/abs/2503.21309, 2025 d

    work page arXiv 2025

  32. [32]

    Cmat: A multi-agent collaboration tuning framework for enhancing small language models

    Xuechen Liang, Yangfan He, Meiling Tao, Yinghui Xia, Jianhui Wang, Tianyu Shi, Jun Wang, and JingSong Yang. Cmat: A multi-agent collaboration tuning framework for enhancing small language models. arXiv preprint arXiv:2404.01663, 2024

    work page arXiv 2024

  33. [33]

    Stock trend prediction based on dynamic hypergraph spatio-temporal network

    Sihao Liao, Liang Xie, Yuanchuang Du, Shengshuang Chen, Hongyang Wan, and Haijiao Xu. Stock trend prediction based on dynamic hypergraph spatio-temporal network. Applied Soft Computing, 154: 0 111329, 2024

    work page 2024

  34. [34]

    Sparsetsf: Modeling long-term time series forecasting with *1k* parameters

    Shengsheng Lin, Weiwei Lin, Wentai Wu, Haojun Chen, and Junjie Yang. Sparsetsf: Modeling long-term time series forecasting with *1k* parameters. In ICML, pp.\ 30211--30226, 2024

    work page 2024

  35. [35]

    Cyclenet: enhancing time series forecasting through modeling periodic patterns

    Shengsheng Lin, Weiwei Lin, Xinyi Hu, Wentai Wu, Ruichao Mo, and Haocheng Zhong. Cyclenet: enhancing time series forecasting through modeling periodic patterns. pp.\ 106315--106345, 2025

    work page 2025

  36. [36]

    Calf: Aligning llms for time series forecasting via cross-modal fine-tuning

    Peiyuan Liu, Hang Guo, Tao Dai, Naiqi Li, Jigang Bao, Xudong Ren, Yong Jiang, and Shu-Tao Xia. Calf: Aligning llms for time series forecasting via cross-modal fine-tuning. In AAAI, volume 39, pp.\ 18915--18923, 2025 a

    work page 2025

  37. [37]

    Timebridge: Non-stationarity matters for long-term time series forecasting

    Peiyuan Liu, Beiliang Wu, Yifan Hu, Naiqi Li, Tao Dai, Jigang Bao, and Shu-tao Xia. Timebridge: Non-stationarity matters for long-term time series forecasting. In ICML, 2025 b

    work page 2025

  38. [38]

    Rethinking irregular time series forecasting: A simple yet effective baseline

    Xvyuan Liu, Xiangfei Qiu, Xingjian Wu, Zhengyu Li, Chenjuan Guo, Jilin Hu, and Bin Yang. Rethinking irregular time series forecasting: A simple yet effective baseline. arXiv preprint arXiv:2505.11250, 2025 c

    work page arXiv 2025

  39. [39]

    Tf-icon: Diffusion-based training-free cross-domain image composition

    Shilin Lu, Yanzhu Liu, and Adams Wai-Kin Kong. Tf-icon: Diffusion-based training-free cross-domain image composition. In ICCV, pp.\ 2294--2305, 2023

    work page 2023

  40. [40]

    Mace: Mass concept erasure in diffusion models

    Shilin Lu, Zilan Wang, Leyang Li, Yanzhu Liu, and Adams Wai-Kin Kong. Mace: Mass concept erasure in diffusion models. In CVPR, pp.\ 6430--6440, 2024 a

    work page 2024

  41. [41]

    Robust watermarking using generative priors against image editing: From benchmarking to advances

    Shilin Lu, Zihan Zhou, Jiayou Lu, Yuanzhi Zhu, and Adams Wai-Kin Kong. Robust watermarking using generative priors against image editing: From benchmarking to advances. arXiv preprint arXiv:2410.18775, 2024 b

    work page arXiv 2024

  42. [42]

    Stock trends prediction based on hypergraph modeling clustering algorithm

    Yongen Luo, Jicheng Hu, Xiaofeng Wei, Dongjian Fang, and Heng Shao. Stock trends prediction based on hypergraph modeling clustering algorithm. In PIC, pp.\ 27--31, 2014

    work page 2014

  43. [43]

    Stock market's price movement prediction with lstm neural networks

    David MQ Nelson, Adriano CM Pereira, and Renato A De Oliveira. Stock market's price movement prediction with lstm neural networks. In IJCNN, pp.\ 1419--1426, 2017

    work page 2017

  44. [44]

    Langtime: A language-guided unified model for time series forecasting with proximal policy optimization

    Wenzhe Niu, Zongxia Xie, Yanru Sun, Wei He, Man Xu, and Chao Hao. Langtime: A language-guided unified model for time series forecasting with proximal policy optimization. In ICML, 2025

    work page 2025

  45. [45]

    Effects of global financial crisis on network structure in a local stock market

    Ashadun Nobi, Seong Eun Maeng, Gyeong Gyun Ha, and Jae Woo Lee. Effects of global financial crisis on network structure in a local stock market. Physica A: Statistical Mechanics and its Applications, 407: 0 135--143, 2014

    work page 2014

  46. [46]

    Decision support system for stock trading using multiple indicators decision tree

    FX Satriyo D Nugroho, Teguh Bharata Adji, and Silmi Fauziati. Decision support system for stock trading using multiple indicators decision tree. In ICITACEE, pp.\ 291--296, 2014

    work page 2014

  47. [47]

    Strategic financial decision-making in sustainable energy investments: Leveraging big data for maximum impact

    Omowonuola Ireoluwapo Kehinde Olanrewaju, Gideon Oluseyi Daramola, and Darlington Eze Ekechukwu. Strategic financial decision-making in sustainable energy investments: Leveraging big data for maximum impact. World Journal of Advanced Research and Reviews, 22 0 (3): 0 564--573, 2024

    work page 2024

  48. [48]

    Yang, Zachary DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fang, Junjie Bai, and Soumith Chintala

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas K \" o pf, Edward Z. Yang, Zachary DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fang, Junjie Bai, and Soumith Chintala. Pytorch: An imperative style, hi...

    work page 2019

  49. [49]

    A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction

    Yao Qin, Dongjin Song, Haifeng Chen, Wei Cheng, Guofei Jiang, and Garrison Cottrell. A dual-stage attention-based recurrent neural network for time series prediction. arXiv preprint arXiv:1704.02971, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  50. [50]

    Jensen, Zhenli Sheng, and Bin Yang

    Xiangfei Qiu, Jilin Hu, Lekui Zhou, Xingjian Wu, Junyang Du, Buang Zhang, Chenjuan Guo, Aoying Zhou, Christian S. Jensen, Zhenli Sheng, and Bin Yang. TFB : Towards comprehensive and fair benchmarking of time series forecasting methods. In Proc. VLDB Endow. , pp.\ 2363--2377, 2024

    work page 2024

  51. [51]

    A comprehensive survey of deep learning for multivariate time series forecasting: A channel strategy perspective

    Xiangfei Qiu, Hanyin Cheng, Xingjian Wu, Jilin Hu, and Chenjuan Guo. A comprehensive survey of deep learning for multivariate time series forecasting: A channel strategy perspective. arXiv preprint arXiv:2502.10721, 2025 a

    work page arXiv 2025

  52. [52]

    Jensen, and Bin Yang

    Xiangfei Qiu, Xiuwen Li, Ruiyang Pang, Zhicheng Pan, Xingjian Wu, Liu Yang, Jilin Hu, Yang Shu, Xuesong Lu, Chengcheng Yang, Chenjuan Guo, Aoying Zhou, Christian S. Jensen, and Bin Yang. EasyTime : Time series forecasting made easy. In ICDE, 2025 b

    work page 2025

  53. [53]

    Jensen, and Bin Yang

    Xiangfei Qiu, Zhe Li, Wanghui Qiu, Shiyan Hu, Lekui Zhou, Xingjian Wu, Zhengyu Li, Chenjuan Guo, Aoying Zhou, Zhenli Sheng, Jilin Hu, Christian S. Jensen, and Bin Yang. Tab: Unified benchmarking of time series anomaly detection methods. In Proc. VLDB Endow. , pp.\ 2775--2789, 2025 c

    work page 2025

  54. [54]

    DUET : Dual clustering enhanced multivariate time series forecasting

    Xiangfei Qiu, Xingjian Wu, Yan Lin, Chenjuan Guo, Jilin Hu, and Bin Yang. DUET : Dual clustering enhanced multivariate time series forecasting. In SIGKDD, pp.\ 1185--1196, 2025 d

    work page 2025

  55. [55]

    DAG: A Dual Correlation Network for Time Series Forecasting with Exogenous Variables

    Xiangfei Qiu, Yuhan Zhu, Zhengyu Li, Hanyin Cheng, Xingjian Wu, Chenjuan Guo, Bin Yang, and Jilin Hu. Dag: A dual causal network for time series forecasting with exogenous variables. arXiv preprint arXiv:2509.14933, 2025 e

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  56. [56]

    A comprehensive energy management application method considering smart home occupant behavior using iot and real big data

    S Saba Rafiei, Mahdi S Naderi, and Mehrdad Abedi. A comprehensive energy management application method considering smart home occupant behavior using iot and real big data. arXiv preprint arXiv:2502.06052, 2025

    work page arXiv 2025

  57. [57]

    Predicting prices of stock market using gated recurrent units (grus) neural networks

    Mohammad Obaidur Rahman, Md Sabir Hossain, Ta-Seen Junaid, Md Shafiul Alam Forhad, and Muhammad Kamal Hossen. Predicting prices of stock market using gated recurrent units (grus) neural networks. International Journal of Computer Science and Network Security, 19 0 (1): 0 213--222, 2019

    work page 2019

  58. [58]

    Stock market prediction and portfolio selection models: a survey

    Akhter Mohiuddin Rather, VN Sastry, and Arun Agarwal. Stock market prediction and portfolio selection models: a survey. Opsearch, 54: 0 558--579, 2017

    work page 2017

  59. [59]

    Spatiotemporal hypergraph convolution network for stock movement forecasting

    Ramit Sawhney, Shivam Agarwal, Arnav Wadhwa, and Rajiv Ratn Shah. Spatiotemporal hypergraph convolution network for stock movement forecasting. In ICDM, pp.\ 482--491, 2020

    work page 2020

  60. [60]

    Stock selection via spatiotemporal hypergraph attention network: A learning to rank approach

    Ramit Sawhney, Shivam Agarwal, Arnav Wadhwa, Tyler Derr, and Rajiv Ratn Shah. Stock selection via spatiotemporal hypergraph attention network: A learning to rank approach. In AAAI, pp.\ 497--504, 2021

    work page 2021

  61. [61]

    The sharpe ratio

    William F Sharpe. The sharpe ratio. Journal of portfolio management, 21 0 (1): 0 49--58, 1994

    work page 1994

  62. [62]

    Learning pattern-specific experts for time series forecasting under patch-level distribution shift

    Yanru Sun, Zongxia Xie, Emadeldeen Eldele, Dongyue Chen, Qinghua Hu, and Min Wu. Learning pattern-specific experts for time series forecasting under patch-level distribution shift. arXiv preprint arXiv:2410.09836, 2024

    work page arXiv 2024

  63. [63]

    Hierarchical classification auxiliary network for time series forecasting

    Yanru Sun, Zongxia Xie, Dongyue Chen, Emadeldeen Eldele, and Qinghua Hu. Hierarchical classification auxiliary network for time series forecasting. In AAAI, volume 39, pp.\ 20743--20751, 2025 a

    work page 2025

  64. [64]

    Ppgf: Probability pattern-guided time series forecasting

    Yanru Sun, Zongxia Xie, Haoyu Xing, Hualong Yu, and Qinghua Hu. Ppgf: Probability pattern-guided time series forecasting. IEEE Transactions on Neural Networks and Learning Systems, 2025 b

    work page 2025

  65. [65]

    Forecasting gold prices with mlp neural networks: A machine learning approach

    Arash Tashakkori, Mohammad Talebzadeh, Fatemeh Salboukh, and Lochan Deshmukh. Forecasting gold prices with mlp neural networks: A machine learning approach. IJSEA, pp.\ 13--20, 2024

    work page 2024

  66. [66]

    Graph Attention Networks

    Petar Veli c kovi \'c , Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. Graph attention networks. arXiv preprint arXiv:1710.10903, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  67. [67]

    Unitmge: Uniform text-motion generation and editing model via diffusion

    Ruoyu Wang, Yangfan He, Tengjiao Sun, Xiang Li, and Tianyu Shi. Unitmge: Uniform text-motion generation and editing model via diffusion. In WACV, pp.\ 6104--6114, 2025

    work page 2025

  68. [68]

    K ^2 VAE : A koopman-kalman enhanced variational autoencoder for probabilistic time series forecasting

    Xingjian Wu, Xiangfei Qiu, Hongfan Gao, Jilin Hu, Bin Yang, and Chenjuan Guo. K ^2 VAE : A koopman-kalman enhanced variational autoencoder for probabilistic time series forecasting. In ICML, 2025 a

    work page 2025

  69. [69]

    CATCH : Channel-aware multivariate time series anomaly detection via frequency patching

    Xingjian Wu, Xiangfei Qiu, Zhengyu Li, Yihang Wang, Jilin Hu, Chenjuan Guo, Hui Xiong, and Bin Yang. CATCH : Channel-aware multivariate time series anomaly detection via frequency patching. In ICLR, 2025 b

    work page 2025

  70. [70]

    Xinle Wu, Xingjian Wu, Bin Yang, Lekui Zhou, Chenjuan Guo, Xiangfei Qiu, Jilin Hu, Zhenli Sheng, and Christian S. Jensen. AutoCTS++ : zero-shot joint neural architecture and hyperparameter search for correlated time series forecasting. VLDB J. , 33 0 (5): 0 1743--1770, 2024

    work page 2024

  71. [71]

    Temporal and heterogeneous graph neural network for financial time series prediction

    Sheng Xiang, Dawei Cheng, Chencheng Shang, Ying Zhang, and Yuqi Liang. Temporal and heterogeneous graph neural network for financial time series prediction. In CIKM, pp.\ 3584--3593, 2022

    work page 2022

  72. [72]

    Wcdt: World-centric diffusion transformer for traffic scene generation

    Chen Yang, Yangfan He, Aaron Xuxiang Tian, Dong Chen, Jianhui Wang, Tianyu Shi, Arsalan Heydarian, and Pei Liu. Wcdt: World-centric diffusion transformer for traffic scene generation. arXiv preprint arXiv:2404.02082, 2024

    work page arXiv 2024

  73. [73]

    A novel forecasting method based on multi-order fuzzy time series and technical analysis

    Furong Ye, Liming Zhang, Defu Zhang, Hamido Fujita, and Zhiguo Gong. A novel forecasting method based on multi-order fuzzy time series and technical analysis. Information Sciences, 367: 0 41--57, 2016

    work page 2016

  74. [74]

    Are transformers effective for time series forecasting? In AAAI, pp.\ 11121--11128, 2023

    Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. Are transformers effective for time series forecasting? In AAAI, pp.\ 11121--11128, 2023

    work page 2023

  75. [75]

    Getting your big data priorities straight: a demonstration of priority-based qos using social-network-driven stock recommendation

    Rui Zhang, Reshu Jain, Prasenjit Sarkar, and Lukas Rupprecht. Getting your big data priorities straight: a demonstration of priority-based qos using social-network-driven stock recommendation. In Proc. VLDB Endow. , pp.\ 1665--1668, 2014

    work page 2014

  76. [76]

    Relational temporal graph convolutional networks for ranking-based stock prediction

    Zetao Zheng, Jie Shao, Jia Zhu, and Heng Tao Shen. Relational temporal graph convolutional networks for ranking-based stock prediction. In ICDE, pp.\ 123--136, 2023

    work page 2023

  77. [77]

    Reagent-v: A reward-driven multi-agent framework for video understanding

    Yiyang Zhou, Yangfan He, Yaofeng Su, Siwei Han, Joel Jang, Gedas Bertasius, Mohit Bansal, and Huaxiu Yao. Reagent-v: A reward-driven multi-agent framework for video understanding. arXiv preprint arXiv:2506.01300, 2025

    work page arXiv 2025

  78. [78]

    Human-centric reward optimization for reinforcement learning-based automated driving using large language models

    Ziqi Zhou, Jingyue Zhang, Jingyuan Zhang, Yangfan He, Boyue Wang, Tianyu Shi, and Alaa Khamis. Human-centric reward optimization for reinforcement learning-based automated driving using large language models. arXiv preprint arXiv:2405.04135, 2024

    work page arXiv 2024

  79. [79]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

    work page

  80. [80]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should not add it explicitly Type <Return> for now, but then later remove the command n...

    work page

Showing first 80 references.