pith. sign in

arxiv: 2605.09395 · v2 · pith:6PLHLFVTnew · submitted 2026-05-10 · 💻 cs.AI · cs.LG· cs.MA· cs.MM

Empowering VLMs for Few-Shot Multimodal Time Series Classification via Tailored Agentic Reasoning

Lin Li , Jiawei Huang , Qihao Quan , Dan Li , Boxin Li , Xiao Zhang , Erli Meng , Wenjie Feng
show 2 more authors
Jian Lou See-Kiong Ng
This is my paper

Pith reviewed 2026-05-20 22:43 UTC · model grok-4.3

classification 💻 cs.AI cs.LGcs.MAcs.MM
keywords few-shot classificationmultimodal time seriesvision language modelsagentic reasoningknowledge bankreflective agentstime series analysis
0
0 comments X

The pith

MarsTSC uses three agent roles to refine a knowledge bank and boost few-shot time series classification in vision-language models.

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

The paper introduces MarsTSC as a framework for few-shot multimodal time series classification that employs agentic reasoning to create a self-evolving knowledge bank. Three agents collaborate: the generator performs classification with reasoning, the reflector identifies errors and highlights missed temporal features, and the modifier updates the knowledge bank to avoid collapse. A test-time update strategy allows ongoing refinement to handle limited data and distribution changes. Experiments on 12 benchmarks with 6 VLM backbones show gains over classical and foundation model baselines along with interpretable explanations for each decision. A reader would care because this makes powerful models usable when labeled examples are scarce.

Core claim

MarsTSC delivers substantial and consistent performance gains across 6 VLM backbones, outperforming both classical and foundation model-based time series baselines under few-shot conditions, while producing interpretable rationales that ground each classification decision in human-readable feature evidence.

What carries the argument

The self-evolving knowledge bank iteratively refined via reflective agentic reasoning with generator, reflector, and modifier agents.

If this is right

  • Outperforms baselines on 12 mainstream time series benchmarks in few-shot settings
  • Works across 6 different VLM backbones with consistent improvements
  • Generates human-readable rationales explaining classifications based on temporal features
  • Uses test-time updates to reduce few-shot bias and distribution shift effects

Where Pith is reading between the lines

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

  • The approach might generalize to other data modalities or tasks requiring few-shot adaptation.
  • Interpretable rationales could support applications where decision transparency is required.
  • Continuous knowledge bank updates may enable better handling of evolving data streams in practice.

Load-bearing premise

The reflector agent can reliably diagnose reasoning errors and produce insights that target specific temporal features the generator missed, so the modifier can make useful updates.

What would settle it

Running the system without the reflector and modifier agents and observing whether performance and interpretability still improve would test if the full agentic loop is necessary.

Figures

Figures reproduced from arXiv: 2605.09395 by Boxin Li, Dan Li, Erli Meng, Jian Lou, Jiawei Huang, Lin Li, Qihao Quan, See-kiong Ng, Wenjie Feng, Xiao Zhang.

Figure 1
Figure 1. Figure 1: Overview of our MarsTSC framework. It comprises three key stages: Warm-up Stage, Training Stage and Testing Stage. sophisticated designs, including TimesNet[46], MultiRocket[41], Autoformer[47], PatchTST[33]. Recent progress in LLMs has fueled increasing interest in adopt￾ing large models for time series tasks. Early attempts includes adapt￾ing pretrained LLMs by fine-tuning on abundant datasets using text… view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of knowledge bank design. Intra-class Feature Initialization. We construct intra-class fea￾ture descriptors to capture prototypical characteristics of samples for each class, covering both the numeric and imaged time series modalities and organizing their complementary features jointly. Specifically, we leverage the cross-modal alignment capabilities of VLMs to analyze the visual representatio… view at source ↗
Figure 3
Figure 3. Figure 3: Comparison with Increasing K-shot Baselines [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: A case about train refinement on success reasoning. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: A case about train refinement on erroneous rea [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: left: Comparison of mean accuracy and standard [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Token count of the knowledge bank over the course [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Classification accuracy with different base VLM models [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Two inference paths that were successful in predict [PITH_FULL_IMAGE:figures/full_fig_p017_9.png] view at source ↗
Figure 12
Figure 12. Figure 12: A case about self-recheck in test time. Pass-2 rea [PITH_FULL_IMAGE:figures/full_fig_p018_12.png] view at source ↗
read the original abstract

In this paper, we propose the first VL$\underline{\textbf{M}}$ $\underline{\textbf{a}}$gentic $\underline{\textbf{r}}$easoning framework for few-$\underline{\textbf{s}}$hot multimodal $\underline{\textbf{T}}$ime $\underline{\textbf{S}}$eries $\underline{\textbf{C}}$lassification ($\textbf{MarsTSC}$), which introduces a self-evolving knowledge bank as a dynamic context iteratively refined via reflective agentic reasoning. The framework comprises three collaborative roles: i) Generator conducts reliable classification via reasoning; ii) Reflector diagnoses the root causes of reasoning errors to yield discriminative insights targeting the temporal features overlooked by Generator; iii) Modifier applies verified updates to the knowledge bank to prevent context collapse. We further introduce a test-time update strategy to enable cautious, continuous knowledge bank refinement to mitigate few-shot bias and distribution shift. Extensive experiments across 12 mainstream time series benchmarks demonstrate that $\textbf{MarsTSC}$ delivers substantial and consistent performance gains across 6 VLM backbones, outperforming both classical and foundation model-based time series baselines under few-shot conditions, while producing interpretable rationales that ground each classification decision in human-readable feature evidence.

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 MarsTSC, the first VLM agentic reasoning framework for few-shot multimodal time series classification. It introduces a self-evolving knowledge bank refined iteratively through reflective agentic reasoning involving three roles: Generator for classification, Reflector for diagnosing reasoning errors and providing insights on overlooked temporal features, and Modifier for updating the knowledge bank. A test-time update strategy is also presented to handle few-shot bias and distribution shift. The authors report substantial and consistent performance gains across 12 time series benchmarks and 6 VLM backbones, outperforming classical and foundation model baselines, along with interpretable rationales.

Significance. If the empirical results hold and the agentic components are shown to be responsible for the gains, this could represent a meaningful advance in applying VLMs to time series tasks by improving few-shot performance and providing human-readable explanations. The tailored agent roles and self-evolving knowledge bank address important limitations in standard prompting approaches for temporal data.

major comments (2)
  1. The central claim attributes substantial gains to the three-role agentic loop in which the Reflector diagnoses root causes of generator errors and yields insights specifically targeting overlooked temporal features (trends, periodicity, phase). For this to explain the reported outperformance over classical and foundation-model baselines under few-shot conditions, the Reflector must produce actionable, time-series-specific corrections. The framework description provides no quantitative metric (e.g., reflector diagnostic accuracy against human labels on misclassified samples) or ablation that isolates the reflector-modifier pair from generic chain-of-thought or self-refinement prompting.
  2. Experiments section: While performance gains are asserted on 12 benchmarks across 6 VLM backbones, the manuscript lacks targeted ablations for the Reflector and Modifier components. Without these, it is impossible to confirm that gains exceed what would be obtained from standard VLM prompting or self-refinement alone, weakening the attribution to the proposed agentic reasoning.
minor comments (2)
  1. Abstract: Consider adding one or two concrete accuracy numbers or relative improvements to give readers an immediate sense of effect size.
  2. Notation: Ensure consistent expansion of acronyms such as VLM on first use in each major section.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on strengthening the empirical validation of the agentic components. We address each major comment below and will revise the manuscript accordingly to better isolate the contributions of the proposed framework.

read point-by-point responses

  1. Referee: The central claim attributes substantial gains to the three-role agentic loop in which the Reflector diagnoses root causes of generator errors and yields insights specifically targeting overlooked temporal features (trends, periodicity, phase). For this to explain the reported outperformance over classical and foundation-model baselines under few-shot conditions, the Reflector must produce actionable, time-series-specific corrections. The framework description provides no quantitative metric (e.g., reflector diagnostic accuracy against human labels on misclassified samples) or ablation that isolates the reflector-modifier pair from generic chain-of-thought or self-refinement prompting.

    Authors: We agree that a dedicated quantitative evaluation of the Reflector's diagnostic quality would strengthen the central claim. The current manuscript provides qualitative case studies in the appendix illustrating how the Reflector identifies overlooked temporal features such as phase shifts and periodicity that the Generator misses, leading to knowledge bank updates. However, we did not include a human-labeled accuracy metric for these diagnoses. We will add an ablation in the revised version that replaces the Reflector-Modifier pair with generic chain-of-thought self-refinement (without the tailored temporal-feature focus) and report the resulting performance drop on the same 12 benchmarks. This will help attribute gains specifically to the proposed roles rather than generic refinement. revision: yes

  2. Referee: Experiments section: While performance gains are asserted on 12 benchmarks across 6 VLM backbones, the manuscript lacks targeted ablations for the Reflector and Modifier components. Without these, it is impossible to confirm that gains exceed what would be obtained from standard VLM prompting or self-refinement alone, weakening the attribution to the proposed agentic reasoning.

    Authors: We acknowledge the absence of component-specific ablations in the submitted manuscript. The reported results compare MarsTSC against classical time-series methods and foundation-model baselines under identical few-shot settings, but do not directly ablate the Reflector or Modifier in isolation. In the revision we will insert a new subsection with targeted ablations: (1) MarsTSC without Reflector (using only Generator + standard prompting), (2) MarsTSC without Modifier (updates applied without verification), and (3) a generic self-refinement baseline. These will be run across the same 6 VLM backbones and 12 datasets to quantify the incremental benefit of the tailored agentic loop. revision: yes

Circularity Check

0 steps flagged

No significant circularity; framework is independently specified

full rationale

The paper defines MarsTSC as a novel three-role agentic framework (Generator for classification, Reflector for diagnosing temporal oversights, Modifier for knowledge-bank updates) plus a test-time refinement strategy. These components are introduced via explicit role descriptions and update rules rather than any equations, fitted parameters, or self-referential definitions. Performance claims rest on external benchmark experiments across 12 datasets and 6 VLMs, not on reductions to the framework's own inputs. No load-bearing self-citations, uniqueness theorems, or ansatzes appear in the derivation chain that would collapse the central claims back to prior author work or internal fits. The overall structure remains self-contained against external evaluation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 2 invented entities

The abstract introduces several new components whose effectiveness rests on unstated assumptions about VLM reasoning capabilities and the ability of reflective diagnosis to improve temporal feature detection.

axioms (1)
  • domain assumption Vision-language models possess sufficient reasoning ability to perform reliable classification on multimodal time series when guided by structured agent roles.
    Invoked by the generator role and the overall framework design.
invented entities (2)
  • self-evolving knowledge bank no independent evidence
    purpose: Dynamic context store that is iteratively refined to prevent context collapse and mitigate few-shot bias.
    Core new component of the framework with no independent evidence provided.
  • Reflector agent no independent evidence
    purpose: Diagnoses root causes of classification errors to extract overlooked temporal features.
    New role introduced without prior validation in the abstract.

pith-pipeline@v0.9.0 · 5779 in / 1430 out tokens · 43494 ms · 2026-05-20T22:43:37.767018+00:00 · methodology

Review history (2 revisions) →

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

71 extracted references · 71 canonical work pages · 8 internal anchors

  1. [1]

    Harika Abburi, Tanya Chaudhary, Haider Ilyas, Lakshmi Manne, Deepak Mit- tal, Don Williams, Derek Snaidauf, Edward Bowen, and Balaji Veeramani

    work page

  2. [2]

    arXiv preprint arXiv:2309.17001 (2023)

    A closer look at bearing fault classification approaches. arXiv preprint arXiv:2309.17001 (2023)

    work page arXiv 2023

  3. [3]

    Yihao Ang, Yifan Bao, Lei Jiang, Jiajie Tao, Anthony KH Tung, Lukasz Szpruch, and Hao Ni. 2025. Structured Agentic Workflows for Financial Time-Series Modeling with LLMs and Reflective Feedback. arXiv preprint arXiv:2508.13915 (2025)

    work page arXiv 2025

  4. [4]

    Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen, Xiyuan Zhang, Pedro Mercado, Huibin Shen, Oleksandr Shchur, Syama Sundar Rangapuram, Sebas- tian Pineda Arango, Shubham Kapoor, et al. 2024. Chronos: Learning the language of time series. arXiv preprint arXiv:2403.07815 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  5. [5]

    Anthony Bagnall, Hoang Anh Dau, Jason Lines, Michael Flynn, James Large, Aaron Bostrom, Paul Southam, and Eamonn Keogh. 2018. The UEA multivariate time series classification archive, 2018. arXiv preprint arXiv:1811.00075 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  6. [6]

    Shelly Bensal, Umar Jamil, Christopher Bryant, Melisa Russak, Kiran Kam- ble, Dmytro Mozolevskyi, Muayad Ali, and Waseem AlShikh. 2025. Reflect, retry, reward: Self-improving llms via reinforcement learning. arXiv preprint arXiv:2505.24726 (2025)

    work page arXiv 2025

  7. [7]

    Leo Breiman. 2001. Random forests. Machine learning 45, 1 (2001), 5–32

    work page 2001

  8. [8]

    Sukru Selim Calik, Andac Akyuz, Zeynep Hilal Kilimci, and Kerem Colak. 2025. Explainable-AI powered stock price prediction using time series transformers: A Case Study on BIST100. arXiv preprint arXiv:2506.06345 (2025)

    work page arXiv 2025

  9. [9]

    Ngai Hang Chan. 2004. Time series: applications to finance. John Wiley & Sons

    work page 2004

  10. [10]

    Ching Chang, Wei-Yao Wang, Wen-Chih Peng, and Tien-Fu Chen. 2025. Llm4ts: Aligning pre-trained llms as data-efficient time-series forecasters. ACM Transactions on Intelligent Systems and Technology 16, 3 (2025), 1–20

    work page 2025

  11. [11]

    Zhe Chen, Jiannan Wu, Wenhai Wang, Weijie Su, Guo Chen, Sen Xing, Muyan Zhong, Qinglong Zhang, Xizhou Zhu, Lewei Lu, et al . 2024. Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 24185–24198

    work page 2024

  12. [12]

    Corinna Cortes and Vladimir Vapnik. 1995. Support-vector networks. Machine learning 20, 3 (1995), 273–297

    work page 1995

  13. [13]

    Thomas Cover and Peter Hart. 1967. Nearest neighbor pattern classification. IEEE transactions on information theory 13, 1 (1967), 21–27

    work page 1967

  14. [14]

    Mayank Daswani, Mathias MJ Bellaiche, Marc Wilson, Desislav Ivanov, Mikhail Papkov, Eva Schnider, Jing Tang, Kay Lamerigts, Gabriela Botea, Michael A Sanchez, et al . 2024. Plots unlock time-series understanding in multimodal models. arXiv preprint arXiv:2410.02637 (2024)

    work page arXiv 2024

  15. [15]

    Hoang Anh Dau, Anthony Bagnall, Kaveh Kamgar, Chin-Chia Michael Yeh, Yan Zhu, Shaghayegh Gharghabi, Chotirat Ann Ratanamahatana, and Eamonn Keogh

    work page

  16. [16]

    IEEE/CAA Journal of Automatica Sinica 6, 6 (2019), 1293–1305

    The UCR time series archive. IEEE/CAA Journal of Automatica Sinica 6, 6 (2019), 1293–1305

    work page 2019

  17. [17]

    Shengdong Du, Tianrui Li, Yan Yang, and Shi-Jinn Horng. 2019. Deep air qual- ity forecasting using hybrid deep learning framework. IEEE Transactions on Knowledge and Data Engineering 33, 6 (2019), 2412–2424

    work page 2019

  18. [18]

    Mojtaba A Farahani, MR McCormick, Ramy Harik, and Thorsten Wuest. 2025. Time-series classification in smart manufacturing systems: An experimen- tal evaluation of state-of-the-art machine learning algorithms. Robotics and Computer-Integrated Manufacturing 91 (2025), 102839

    work page 2025

  19. [19]

    Google DeepMind. 2026. Gemini 3.1 Pro Model Card. https://deepmind.google/ models/model-cards/gemini-3-1-pro/

    work page 2026

  20. [20]

    Mononito Goswami, Konrad Szafer, Arjun Choudhry, Yifu Cai, Shuo Li, and Artur Dubrawski. 2024. MOMENT: A Family of Open Time-series Foundation Models. In International Conference on Machine Learning. PMLR, 16115–16152

    work page 2024

  21. [21]

    Xinyu Huang, Jun Tang, and Yongming Shen. 2024. Long time series of ocean wave prediction based on PatchTST model. Ocean Engineering 301 (2024), 117572

    work page 2024

  22. [22]

    Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. 2023. Survey of hallucination in natural language generation. ACM computing surveys 55, 12 (2023), 1–38

    work page 2023

  23. [23]

    Yushan Jiang, Kanghui Ning, Zijie Pan, Xuyang Shen, Jingchao Ni, Wenchao Yu, Anderson Schneider, Haifeng Chen, Yuriy Nevmyvaka, and Dongjin Song. 2025. Multi-modal time series analysis: A tutorial and survey. In Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2. 6043–6053

    work page 2025

  24. [24]

    Yushan Jiang, Wenchao Yu, Geon Lee, Dongjin Song, Kijung Shin, Wei Cheng, Yanchi Liu, and Haifeng Chen. 2025. Timexl: Explainable multi-modal time series prediction with llm-in-the-loop. arXiv preprint arXiv:2503.01013 (2025)

    work page arXiv 2025

  25. [25]

    Mohammad Ali Labbaf Khaniki, Alireza Golkarieh, Houman Nouri, and Moham- mad Manthouri. 2024. Enhanced fault detection and cause identification using integrated attention mechanism. arXiv preprint arXiv:2408.00033 (2024)

    work page arXiv 2024

  26. [26]

    Peiwen Li, Xin Wang, Zeyang Zhang, Yuan Meng, Fang Shen, Yue Li, Jialong Wang, Yang Li, and Wenwu Zhu. 2024. Realtcd: Temporal causal discovery from interventional data with large language model. In Proceedings of the 33rd ACM international conference on information and knowledge management. 4669– 4677

    work page 2024

  27. [27]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023. Visual in- struction tuning. Advances in neural information processing systems 36 (2023), 34892–34916

    work page 2023

  28. [28]

    Xu Liu, Juncheng Liu, Gerald Woo, Taha Aksu, Yuxuan Liang, Roger Zimmer- mann, Chenghao Liu, Silvio Savarese, Caiming Xiong, and Doyen Sahoo. 2024. Moirai-moe: Empowering time series foundation models with sparse mixture of experts. arXiv preprint arXiv:2410.10469 (2024)

    work page arXiv 2024

  29. [29]

    Matthew Middlehurst, James Large, Michael Flynn, Jason Lines, Aaron Bostrom, and Anthony Bagnall. 2021. HIVE-COTE 2.0: a new meta ensemble for time series classification. Machine Learning 110, 11 (2021), 3211–3243

    work page 2021

  30. [30]

    Mukaffi Bin Moin, Fatema Tuj Johora Faria, Swarnajit Saha, Busra Kamal Rafa, and Mohammad Shafiul Alam. 2024. Exploring explainable ai techniques for improved interpretability in lung and colon cancer classification. InInternational Conference on Computing and Communication Networks. Springer, 1–11

    work page 2024

  31. [31]

    Moonshot AI. 2026. Kimi K2.5 Quickstart. https://platform.kimi.ai/docs/guide/ kimi-k2-5-quickstart

    work page 2026

  32. [32]

    Mohammad Amin Morid, Olivia R Liu Sheng, and Joseph Dunbar. 2023. Time series prediction using deep learning methods in healthcare. ACM Transactions on Management Information Systems 14, 1 (2023), 1–29

    work page 2023

  33. [33]

    Ozan Baris Mulayim, Pengrui Quan, Liying Han, Xiaomin Ouyang, Dezhi Hong, Mario Bergés, and Mani Srivastava. 2025. Can Time-Series Foundation Models Perform Building Energy Management Tasks? arXiv preprint arXiv:2506.11250 (2025)

    work page arXiv 2025

  34. [34]

    Jingchao Ni, Ziming Zhao, ChengAo Shen, Hanghang Tong, Dongjin Song, Wei Cheng, Dongsheng Luo, and Haifeng Chen. 2025. Harnessing vision models for time series analysis: A survey. arXiv preprint arXiv:2502.08869 (2025)

    work page arXiv 2025

  35. [35]

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. 2022. A time series is worth 64 words: Long-term forecasting with transformers. arXiv preprint arXiv:2211.14730 (2022)

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  36. [36]

    OpenAI. 2026. GPT-5 Model. https://developers.openai.com/api/docs/models/ gpt-5

    work page 2026

  37. [37]

    OpenAI. 2026. GPT-5.4 mini Model. https://developers.openai.com/api/docs/ models/gpt-5.4-mini

    work page 2026

  38. [38]

    Jing-Cheng Pang, Pengyuan Wang, Kaiyuan Li, Xiong-Hui Chen, Jiacheng Xu, Zongzhang Zhang, and Yang Yu. 2023. Language model self-improvement by reinforcement learning contemplation. arXiv preprint arXiv:2305.14483 (2023)

    work page arXiv 2023

  39. [39]

    Ross Quinlan

    J. Ross Quinlan. 1986. Induction of decision trees. Machine learning 1, 1 (1986), 81–106

    work page 1986

  40. [40]

    2026.Qwen3.5-397B-A17B

    Qwen Team. 2026.Qwen3.5-397B-A17B. https://huggingface.co/Qwen/Qwen3.5- 397B-A17B

    work page 2026

  41. [41]

    Lisa Schmors, Dominic Gonschorek, Jan Niklas Böhm, Yongrong Qiu, Na Zhou, Dmitry Kobak, Andreas Tolias, Fabian Sinz, Jacob Reimer, Katrin Franke, et al

    work page

  42. [42]

    arXiv preprint arXiv:2506.04906 (2025)

    TRACE: Contrastive learning for multi-trial time-series data in neuroscience. arXiv preprint arXiv:2506.04906 (2025)

    work page internal anchor Pith review arXiv 2025

  43. [43]

    ChengAo Shen, Wenchao Yu, Ziming Zhao, Dongjin Song, Wei Cheng, Haifeng Chen, and Jingchao Ni. 2025. Multi-modal view enhanced large vision models for long-term time series forecasting. arXiv preprint arXiv:2505.24003 (2025)

    work page arXiv 2025

  44. [44]

    Chang Wei Tan, Angus Dempster, Christoph Bergmeir, and Geoffrey I Webb

    work page

  45. [45]

    Data Mining and Knowledge Discovery 36, 5 (2022), 1623–1646

    MultiRocket: multiple pooling operators and transformations for fast and effective time series classification: CW Tan. Data Mining and Knowledge Discovery 36, 5 (2022), 1623–1646

    work page 2022

  46. [46]

    Mingtian Tan, Mike Merrill, Vinayak Gupta, Tim Althoff, and Tom Hartvigsen

    work page

  47. [47]

    Are language models actually useful for time series forecasting? Advances in Neural Information Processing Systems 37 (2024), 60162–60191

    work page 2024

  48. [48]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. Advances in neural information processing systems 30 (2017)

    work page 2017

  49. [49]

    Oriol Vinyals, Charles Blundell, Timothy Lillicrap, Daan Wierstra, et al . 2016. Matching networks for one shot learning. Advances in neural information processing systems 29 (2016)

    work page 2016

  50. [50]

    Jiahao Wang, Mingyue Cheng, Qingyang Mao, Yitong Zhou, Daoyu Wang, Qi Liu, Feiyang Xu, and Xin Li. 2025. Tabletime: Reformulating time series classification as training-free table understanding with large language models. In Proceedings of the 34th ACM International Conference on Information and Knowledge Management. 3009–3019

    work page 2025

  51. [51]

    Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. [n. d.]. TimesNet: Temporal 2D-Variation Modeling for General Time Series Anal- ysis. In The Eleventh International Conference on Learning Representations

    work page

  52. [52]

    Haixu Wu, Jiehui Xu, Jianmin Wang, and Mingsheng Long. 2021. Autoformer: De- composition transformers with auto-correlation for long-term series forecasting. Advances in neural information processing systems 34 (2021), 22419–22430

    work page 2021

  53. [53]

    Wujiang Xu, Zujie Liang, Kai Mei, Hang Gao, Juntao Tan, and Yongfeng Zhang

    work page

  54. [54]

    A-MEM: Agentic Memory for LLM Agents

    A-mem: Agentic memory for llm agents. arXiv preprint arXiv:2502.12110 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  55. [55]

    An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, et al. 2025. Qwen3 technical 10 Empowering VLMs for Few-Shot Multimodal Time Series Classification via Tailored Agentic Reasoning Preprint, 2026, report. arXiv preprint arXiv:2505.09388 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  56. [56]

    Jinning Yang and Wen Shi. 2025. DiagECG: An LLM-Driven Framework for Diagnostic Reasoning via Discretized ECG Tokenization. arXiv preprint arXiv:2508.15338 (2025)

    work page arXiv 2025

  57. [57]

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2022. React: Synergizing reasoning and acting in language models. In The eleventh international conference on learning representations

    work page 2022

  58. [58]

    Wojciech Zaremba, Ilya Sutskever, and Oriol Vinyals. 2014. Recurrent neural network regularization. arXiv preprint arXiv:1409.2329 (2014)

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  59. [59]

    Qizheng Zhang, Changran Hu, Shubhangi Upasani, Boyuan Ma, Fenglu Hong, Vamsidhar Kamanuru, Jay Rainton, Chen Wu, Mengmeng Ji, Hanchen Li, et al

    work page

  60. [60]

    Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models

    Agentic context engineering: Evolving contexts for self-improving language models. arXiv preprint arXiv:2510.04618 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  61. [61]

    Haokun Zhao, Xiang Zhang, Jiaqi Wei, Yiwei Xu, Yuting He, Siqi Sun, and Chenyu You. 2025. Timeseriesscientist: A general-purpose ai agent for time series analysis. arXiv preprint arXiv:2510.01538 (2025)

    work page arXiv 2025

  62. [62]

    Ziming Zhao, ChengAo Shen, Hanghang Tong, Dongjin Song, Zhigang Deng, Qingsong Wen, and Jingchao Ni. 2025. From Images to Signals: Are Large Vision Models Useful for Time Series Analysis? arXiv preprint arXiv:2505.24030 (2025)

    work page arXiv 2025

  63. [63]

    Siru Zhong, Weilin Ruan, Ming Jin, Huan Li, Qingsong Wen, and Yuxuan Liang

    work page

  64. [64]

    arXiv preprint arXiv:2502.04395 (2025)

    Time-vlm: Exploring multimodal vision-language models for augmented time series forecasting. arXiv preprint arXiv:2502.04395 (2025)

    work page arXiv 2025

  65. [65]

    Shu Zhou, Yunyang Xuan, Yuxuan Ao, Xin Wang, Tao Fan, and Hao Wang. 2025. MERIT: Multi-Agent Collaboration for Unsupervised Time Series Representation Learning. In Findings of the Association for Computational Linguistics: ACL

    work page 2025

  66. [66]

    Tian Zhou, Peisong Niu, Liang Sun, Rong Jin, et al . 2023. One fits all: Power general time series analysis by pretrained lm. Advances in neural information processing systems 36 (2023), 43322–43355

    work page 2023

  67. [67]

    Jiaxin Zhuang, Leon Yan, Zhenwei Zhang, Ruiqi Wang, Jiawei Zhang, and Yuantao Gu. 2024. See it, think it, sorted: Large multimodal models are few-shot time series anomaly analyzers. arXiv preprint arXiv:2411.02465 (2024)

    work page arXiv 2024

  68. [68]

    Yufan Zhuang, Chandan Singh, Liyuan Liu, Yelong Shen, Dinghuai Zhang, Jingbo Shang, Jianfeng Gao, and Weizhu Chen. 2026. Test-time Recursive Thinking: Self-Improvement without External Feedback. arXiv preprint arXiv:2602.03094 (2026). 11 Preprint, 2026, Li et al. A Appendix Overview This appendix is organized as follows. We first describe additional imple...

    work page arXiv 2026

  69. [69]

    - section: the section to add the new bullet to

    ADD: Create new bullet points with fresh IDs. - section: the section to add the new bullet to. - content: the new content of the bullet

    work page

  70. [70]

    - target_id: the exact ID of the bullet to modify

    MODIFY: Update an existing bullet point. - target_id: the exact ID of the bullet to modify. - content: the fully updated content of the bullet

    work page

  71. [71]

    - target_id: index of the bullet point you want to remove

    DELETE: Remove an existing bullet point. - target_id: index of the bullet point you want to remove. ### Query Sample: Below is the Query Image to be classified by the predictor.: { visualized time series } C Extended Experiments C.1 Analysis of Few-shot Train Samples To further evaluate the robustness ofMarsTSCunder few-shot sampling randomness and varyin...

    work page arXiv 2026