pith. sign in

arxiv: 2606.19501 · v1 · pith:4VG3YXM4new · submitted 2026-06-17 · 💻 cs.AI · cs.CL· cs.LG· q-fin.RM

DeXposure-Claw: An Agentic System for DeFi Risk Supervision

Aijie Shu , Bowei Chen , Wenbin Wu , Cathy Yi-Hsuan Chen , Fengxiang He This is my paper

Pith reviewed 2026-06-26 20:55 UTC · model grok-4.3

classification 💻 cs.AI cs.CLcs.LGq-fin.RM
keywords DeFi risk supervisionagentic LLM systemsgraph time-series forecastingrisk alertsfalse intervention ratesupervisory ticketsdeterministic monitors
0
0 comments X

The pith

A forecast-grounded agentic system routes LLM decisions through deterministic monitors to generate auditable DeFi supervisory tickets with controlled false-intervention rates.

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

The paper introduces DeXposure-Claw as a system that first uses a graph time-series foundation model to forecast future exposure networks in decentralized finance. Deterministic monitors and stress scenarios then convert those forecasts into typed alerts and attribution signals. Data-health and confidence gates constrain escalation so that the system emits auditable tickets only when evidence meets regulator-aligned criteria. Experiments on five years of weekly real data are presented as full support for the approach. The evaluation uses a six-axis harness that scores decisions against absolute-loss ground truth and an explicit false-intervention rate.

Core claim

DeXposure-Claw routes LLM decisions through a three-part pipeline of DeXposure-FM forecasts, deterministic monitors that produce typed alerts and scenario evidence, and data-health plus confidence gates that limit escalation, yielding auditable supervisory tickets whose performance on five years of weekly real data fully supports the system when measured by a regulator-aligned false-intervention rate.

What carries the argument

The three-part pipeline that converts graph time-series forecasts into typed alerts via deterministic monitors and stress scenarios, then applies data-health and confidence gates before emitting tickets.

Load-bearing premise

The deterministic monitors and stress scenarios turn the forecasts into alerts and signals without introducing systematic bias or missing material risks.

What would settle it

A test on held-out weekly data from a subsequent period where the system's false-intervention rate exceeds the rate observed in the five-year training window or where a documented loss event produces no ticket.

Figures

Figures reproduced from arXiv: 2606.19501 by Aijie Shu, Bowei Chen, Cathy Yi-Hsuan Chen, Fengxiang He, Wenbin Wu.

Figure 1
Figure 1. Figure 1: DeXposure-Claw system overview. A weekly exposure graph [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
read the original abstract

Decentralized finance exposes supervisors to fast-moving, networked credit risks. General-purpose LLM agents fit this setting poorly: they over-read weak evidence and recommend high-stakes interventions, while existing evaluations offer no regulator-aligned way to measure the resulting false alarms. We introduce DeXposure-Claw, a forecast-grounded agentic supervision system that routes LLM decisions through structured evidence: (1) DeXposure-FM, a graph time-series foundation model, forecasts future exposure networks; (2) deterministic monitors and stress scenarios then turn those forecasts into typed alerts, attribution signals, and scenario evidence; and (3) data-health and confidence gates constrain escalation before DeXposure-Claw emits auditable supervisory tickets with rationales. We further develop DeXposure-Bench, a six-axis evaluation harness, whose decision axis scores tickets against a regulator-aligned absolute-loss ground truth and an explicit false-intervention rate. Experiments on five years of weekly real data fully support our system. Code is at https://github.com/EVIEHub/DeXposure-Claw.

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 / 1 minor

Summary. The manuscript introduces DeXposure-Claw, a forecast-grounded agentic system for supervising fast-moving networked credit risks in DeFi. The system routes LLM decisions through three components: (1) DeXposure-FM, a graph time-series foundation model that forecasts future exposure networks; (2) deterministic monitors and stress scenarios that convert forecasts into typed alerts, attribution signals, and scenario evidence; and (3) data-health and confidence gates that constrain escalation before emitting auditable supervisory tickets. It also presents DeXposure-Bench, a six-axis evaluation harness whose decision axis scores outputs against a regulator-aligned absolute-loss ground truth and an explicit false-intervention rate. The central claim is that experiments on five years of weekly real data fully support the system.

Significance. If the empirical results and monitor completeness arguments hold, the work could supply a structured, auditable alternative to general-purpose LLM agents for DeFi supervision by enforcing explicit false-intervention metrics and regulator-aligned ground truth. The public code release at the cited GitHub repository is a clear strength for reproducibility.

major comments (2)
  1. [Abstract] Abstract: the claim that 'Experiments on five years of weekly real data fully support our system' is presented without any quantitative results, error metrics, baseline comparisons, or description of how the absolute-loss ground truth was constructed. This renders the central empirical claim impossible to assess from the supplied text.
  2. [Abstract] Abstract (three-part pipeline paragraph): the deterministic monitors and stress scenarios are described only at the level of converting DeXposure-FM forecasts into typed alerts and attribution signals, with no design details, completeness arguments, ablation studies, or evidence that they capture correlated credit exposures or avoid systematic false negatives in DeFi graphs. This is the load-bearing conversion layer for both the ground truth and the reported false-intervention rates.
minor comments (1)
  1. [Abstract] The abstract would be strengthened by a single sentence summarizing the key quantitative outcomes (e.g., false-intervention rate, decision-axis score) rather than the unqualified assertion of support.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on the abstract. We address each point below and will revise the manuscript accordingly to improve assessability while preserving the concise nature of the abstract.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that 'Experiments on five years of weekly real data fully support our system' is presented without any quantitative results, error metrics, baseline comparisons, or description of how the absolute-loss ground truth was constructed. This renders the central empirical claim impossible to assess from the supplied text.

    Authors: We agree the abstract claim is stated at too high a level. The experiments section of the manuscript reports the quantitative results, error metrics, baselines, and ground-truth construction details. In revision we will add a concise sentence to the abstract summarizing the key metrics (e.g., decision-axis scores and false-intervention rates) and the absolute-loss ground-truth methodology so the claim can be evaluated from the abstract alone. revision: yes

  2. Referee: [Abstract] Abstract (three-part pipeline paragraph): the deterministic monitors and stress scenarios are described only at the level of converting DeXposure-FM forecasts into typed alerts and attribution signals, with no design details, completeness arguments, ablation studies, or evidence that they capture correlated credit exposures or avoid systematic false negatives in DeFi graphs. This is the load-bearing conversion layer for both the ground truth and the reported false-intervention rates.

    Authors: The abstract is intentionally high-level; the design details, completeness arguments, ablation studies, and evidence on correlated exposures and false-negative avoidance appear in Sections 3.2 and 4. To address the concern we will expand the abstract by one sentence that summarizes the monitor architecture and the main empirical evidence on coverage and false-negative behavior, thereby making the conversion layer more visible without duplicating the full technical treatment. revision: yes

Circularity Check

0 steps flagged

No derivation chain; empirical system evaluation is self-contained

full rationale

The paper presents a three-part agentic system (DeXposure-FM forecasts, deterministic monitors/stress scenarios, and gates) whose central claim is that five years of weekly real-data experiments support the system. No equations, fitting procedures, or mathematical derivations appear in the abstract or described pipeline. The evaluation relies on an external regulator-aligned ground truth and explicit false-intervention rates rather than any self-referential reduction of outputs to inputs. Because no load-bearing derivation exists that could reduce to fitted parameters or self-citations by construction, the circularity score is 0 and steps is empty.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract supplies no equations, no fitted parameters, and no explicit axioms or invented entities; the system description rests on unstated assumptions about forecast accuracy and monitor neutrality that are not enumerated.

pith-pipeline@v0.9.1-grok · 5735 in / 1179 out tokens · 25369 ms · 2026-06-26T20:55:47.480348+00:00 · methodology

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

79 extracted references · 18 canonical work pages

  1. [1]

    Towards Verifiability of Total Value Locked (

    Auer, Raphael and Farag. Towards Verifiability of Total Value Locked (. 2025 , url =

    2025

  2. [2]

    Systemic Risk in

    Zhang, Shiyu and Wang, Zining and Zheng, Jin and Cartlidge, John , journal =. Systemic Risk in. 2026 , url =

    2026

  3. [3]

    Measuring DeFi risk , journal =

    Jeremy Bertomeu and Xiumin Martin and Ibrahima Sall , keywords =. Measuring DeFi risk , journal =. 2024 , issn =. doi:10.1016/j.frl.2024.105321 , url =

    work page doi:10.1016/j.frl.2024.105321 2024

  4. [4]

    2025 , eprint=

    DeXposure: A Dataset and Benchmarks for Inter-protocol Credit Exposure in Decentralized Financial Networks , author=. 2025 , eprint=

    2025

  5. [5]

    2025 , eprint=

    Mapping Microscopic and Systemic Risks in TradFi and DeFi: a literature review , author=. 2025 , eprint=

    2025

  6. [6]

    Mathematics , VOLUME =

    Dolfin, Marina and Knopoff, Damian and Limosani, Michele and Xibilia, Maria Gabriella , TITLE =. Mathematics , VOLUME =. 2019 , NUMBER =

    2019

  7. [7]

    Predicting systemic risk in financial systems using Deep Graph Learning , journal =

    Vicente Balmaseda and María Coronado and Gonzalo. Predicting systemic risk in financial systems using Deep Graph Learning , journal =. 2023 , issn =. doi:10.1016/j.iswa.2023.200240 , url =

    work page doi:10.1016/j.iswa.2023.200240 2023

  8. [8]

    2025 , eprint=

    Computing Systemic Risk Measures with Graph Neural Networks , author=. 2025 , eprint=

    2025

  9. [9]

    Do Transformers Really Perform Badly for Graph Representation? , url =

    Ying, Chengxuan and Cai, Tianle and Luo, Shengjie and Zheng, Shuxin and Ke, Guolin and He, Di and Shen, Yanming and Liu, Tie-Yan , booktitle =. Do Transformers Really Perform Badly for Graph Representation? , url =

  10. [10]

    2025 , eprint=

    Turning Tabular Foundation Models into Graph Foundation Models , author=. 2025 , eprint=

    2025

  11. [11]

    2025 , month = oct, url =

    Crypto-assets and decentralised finance: Report on stablecoins, crypto-investment products and multi-function groups , author =. 2025 , month = oct, url =

    2025

  12. [12]

    Federal Reserve Bank of St

    Decentralized Finance: On Blockchain- and Smart Contract-Based Financial Markets , author =. Federal Reserve Bank of St. Louis Review , volume =. 2021 , doi =

    2021

  13. [13]

    2024 , eprint=

    SoK: Decentralized Finance (DeFi) -- Fundamentals, Taxonomy and Risks , author=. 2024 , eprint=

    2024

  14. [14]

    2023 , month = sep, doi =

    Runs and Flights to Safety: Are Stablecoins the New Money Market Funds? , author =. 2023 , month = sep, doi =

    2023

  15. [15]

    2023 , month = nov, url =

    Will the Real Stablecoin Please Stand Up? , author =. 2023 , month = nov, url =

    2023

  16. [16]

    Management Science , volume =

    Systemic Risk in Financial Systems , author =. Management Science , volume =. 2001 , month = feb, doi =

    2001

  17. [17]

    Acemoglu, A

    Acemoglu, Daron and Ozdaglar, Asuman and Tahbaz-Salehi, Alireza , Title =. American Economic Review , Volume =. 2015 , Month =. doi:10.1257/aer.20130456 , URL =

    work page doi:10.1257/aer.20130456 2015

  18. [18]

    Peyton , Title =

    Glasserman, Paul and Young, H. Peyton , Title =. Journal of Economic Literature , Volume =. 2016 , Month =. doi:10.1257/jel.20151228 , URL =

    work page doi:10.1257/jel.20151228 2016

  19. [19]

    Journal of Economic Perspectives , Volume =

    Mullainathan, Sendhil and Spiess, Jann , Title =. Journal of Economic Perspectives , Volume =. 2017 , Month =. doi:10.1257/jep.31.2.87 , URL =

    work page doi:10.1257/jep.31.2.87 2017

  20. [20]

    International Journal of Computer Applications , issue_date =

    Shahzad Qaiser, Ramsha Ali , title =. International Journal of Computer Applications , issue_date =. 2018 , issn =. doi:10.5120/ijca2018917395 , publisher =

    work page doi:10.5120/ijca2018917395 2018

  21. [21]

    Sims , journal =

    Christopher A. Sims , journal =. Macroeconomics and Reality , urldate =

  22. [22]

    Journal of the American Statistical Association , volume =

    James H Stock and Mark W Watson , title =. Journal of the American Statistical Association , volume =. 2002 , publisher =. doi:10.1198/016214502388618960 , URL =

    work page doi:10.1198/016214502388618960 2002

  23. [23]

    Financial Cryptography and Data Security: 25th International Conference, FC 2021, Virtual Event, March 1–5, 2021, Revised Selected Papers, Part I , pages =

    Qin, Kaihua and Zhou, Liyi and Livshits, Benjamin and Gervais, Arthur , title =. Financial Cryptography and Data Security: 25th International Conference, FC 2021, Virtual Event, March 1–5, 2021, Revised Selected Papers, Part I , pages =. 2021 , isbn =. doi:10.1007/978-3-662-64322-8_1 , abstract =

    work page doi:10.1007/978-3-662-64322-8_1 2021

  24. [24]

    2023 , eprint =

    Toxic Liquidation Spirals , author =. 2023 , eprint =. doi:10.48550/arXiv.2212.07306 , url =

    work page doi:10.48550/arxiv.2212.07306 2023

  25. [25]

    2017 , eprint=

    Semi-Supervised Classification with Graph Convolutional Networks , author=. 2017 , eprint=

    2017

  26. [26]

    2018 , eprint=

    Graph Attention Networks , author=. 2018 , eprint=

    2018

  27. [27]

    2020 , eprint=

    Temporal Graph Networks for Deep Learning on Dynamic Graphs , author=. 2020 , eprint=

    2020

  28. [28]

    Scientific Reports , volume =

    DebtRank: Too Central to Fail? Financial Networks, the FED and Systemic Risk , author =. Scientific Reports , volume =. 2012 , month = aug, doi =

    2012

  29. [29]

    Diebold and Kamil Yilmaz , keywords =

    Francis X. Diebold and Kamil Yilmaz , keywords =. Better to give than to receive: Predictive directional measurement of volatility spillovers , journal =. 2012 , note =. doi:10.1016/j.ijforecast.2011.02.006 , url =

    work page doi:10.1016/j.ijforecast.2011.02.006 2012

  30. [30]

    2024 , eprint =

    Chronos: Learning the Language of Time Series , author =. 2024 , eprint =

    2024

  31. [31]

    2024 , eprint=

    Lag-Llama: Towards Foundation Models for Probabilistic Time Series Forecasting , author=. 2024 , eprint=

    2024

  32. [32]

    2024 , eprint =

    A decoder-only foundation model for time-series forecasting , author =. 2024 , eprint =

    2024

  33. [33]

    and Sun, Yiqiao , institution =

    de Bondt, Gabe J. and Sun, Yiqiao , institution =. Enhancing. 2025 , month = jun, address =. doi:10.2866/2788332 , url =

    work page doi:10.2866/2788332 2025

  34. [34]

    BIS Quarterly Review , year =

    Large language models: a primer for economists , author =. BIS Quarterly Review , year =

  35. [35]

    2025 , eprint=

    Macroeconomic Forecasting with Large Language Models , author=. 2025 , eprint=

    2025

  36. [36]

    Nature , year=

    Accurate predictions on small data with a tabular foundation model , author =. Nature , year =. doi:10.1038/s41586-024-08328-6 , url =

    work page doi:10.1038/s41586-024-08328-6

  37. [37]

    52 Ivan Vynyavskyy, Stefan Kitzler, Bernhard Haslhofer, and Aviv Yaish

    David Vidal-Tomás and Antonio Briola and Tomaso Aste , keywords =. FTX’s downfall and Binance’s consolidation: The fragility of centralised digital finance , journal =. 2023 , issn =. doi:10.1016/j.physa.2023.129044 , url =

    work page doi:10.1016/j.physa.2023.129044 2023

  38. [38]

    2022 , doi =

    You, Jiaxuan and Du, Tianyu and Leskovec, Jure , booktitle =. 2022 , doi =

    2022

  39. [39]

    Proceedings of the 3rd International Conference on Learning Representations (ICLR) , year =

    Adam: A Method for Stochastic Optimization , author =. Proceedings of the 3rd International Conference on Learning Representations (ICLR) , year =

  40. [40]

    and Zhang, Hao and Gonzalez, Joseph E

    Zheng, Lianmin and Chiang, Wei-Lin and Sheng, Ying and Zhuang, Siyuan and Wu, Zhanghao and Zhuang, Yonghao and Lin, Zi and Li, Zhuohan and Li, Dacheng and Xing, Eric P. and Zhang, Hao and Gonzalez, Joseph E. and Stoica, Ion , booktitle =. Judging. 2023 , url=

    2023

  41. [41]

    and Bentley, Michael A

    Aspembitova, Ayana T. and Bentley, Michael A. , TITLE =. Entropy , VOLUME =. 2023 , NUMBER =

    2023

  42. [42]

    2024 , note =

    The Rise of Diffusion Models in Time-Series Forecasting , author =. 2024 , note =

    2024

  43. [43]

    Sharpness-aware Minimization for Efficiently Improving Generalization , author=

  44. [44]

    , booktitle =

    Li, Xiang Lisa and Thickstun, John and Gulrajani, Ishaan and Liang, Percy and Hashimoto, Tatsunori B. , booktitle =. Diffusion-. 2022 , url =

    2022

  45. [45]

    Attention is All you Need , url =

    Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob and Jones, Llion and Gomez, Aidan N and Kaiser, ukasz and Polosukhin, Illia , booktitle =. Attention is All you Need , url =

  46. [46]

    2021 , eprint =

    On the Opportunities and Risks of Foundation Models , author =. 2021 , eprint =

    2021

  47. [47]

    Benítez , keywords =

    Christoph Bergmeir and José M. Benítez , keywords =. On the use of cross-validation for time series predictor evaluation , journal =. 2012 , note =. doi:10.1016/j.ins.2011.12.028 , url =

    work page doi:10.1016/j.ins.2011.12.028 2012

  48. [48]

    Early Stopping - But When?

    Prechelt, Lutz. Early Stopping - But When?. Neural Networks: Tricks of the Trade. 1998. doi:10.1007/3-540-49430-8_3

    work page doi:10.1007/3-540-49430-8_3 1998

  49. [49]

    Proceedings of the 30th International Conference on Machine Learning , pages =

    On the difficulty of training recurrent neural networks , author =. Proceedings of the 30th International Conference on Machine Learning , pages =. 2013 , editor =

    2013

  50. [50]

    Federal Reserve Bulletin , year =

    The Herfindahl-Hirschman index , author =. Federal Reserve Bulletin , year =

  51. [51]

    The anatomy of a large-scale hypertextual Web search engine.Computer Networks and ISDN Systems, 30(1–7):107–117, 1998

    Sergey Brin and Lawrence Page , keywords =. The anatomy of a large-scale hypertextual Web search engine , journal =. 1998 , note =. doi:10.1016/S0169-7552(98)00110-X , url =

    work page doi:10.1016/s0169-7552(98)00110-x 1998

  52. [52]

    An introduction to ROC analysis , journal =

    Tom Fawcett , keywords =. An introduction to ROC analysis , journal =. 2006 , note =. doi:10.1016/j.patrec.2005.10.010 , url =

    work page doi:10.1016/j.patrec.2005.10.010 2006

  53. [53]

    doi: 10.1145/1143844.1143874

    Davis, Jesse and Goadrich, Mark , title =. Proceedings of the 23rd International Conference on Machine Learning , pages =. 2006 , isbn =. doi:10.1145/1143844.1143874 , abstract =

    work page doi:10.1145/1143844.1143874 2006

  54. [54]

    arXiv preprint arXiv:2501.00826 , year=

    LLM-Powered Multi-Agent System for Automated Crypto Portfolio Management , author=. arXiv preprint arXiv:2501.00826 , year=

    Pith/arXiv arXiv

  55. [55]

    Can Blindfolded

    Jeon, Joohyoung and Lee, Hongchul , year =. Can Blindfolded. 2603.17692 , archivePrefix =

    arXiv

  56. [56]

    Auditable

    Yao, Duanyi and Jagannath, Siddhartha and Aroso, Baltasar and Krishnan, Vyas and Zhao, Ding , booktitle=. Auditable. 2026 , url=

    2026

  57. [57]

    Know Your Intent: An Autonomous Multi-Perspective

    Mao, Qian'ang and Zhang, Yuxuan and Chen, Jiaman and Zhou, Wenjun and Yan, Jiaqi , year =. Know Your Intent: An Autonomous Multi-Perspective. 2511.15456 , archivePrefix =

    arXiv

  58. [58]

    2026 , eprint =

    An Effective and Cost-Efficient Agentic Framework for Ethereum Smart Contract Auditing , author =. 2026 , eprint =

    2026

  59. [59]

    2026 , eprint =

    Knowdit: Agentic Smart Contract Vulnerability Detection with Auditing Knowledge Summarization , author =. 2026 , eprint =

    2026

  60. [60]

    2510.21272 , archivePrefix =

    Liu, Lu and Zhang, Wuqi and Wei, Lili and Guan, Hao and Tian, Yongqiang and Liu, Yepang and Cheung, Shing-Chi , year =. 2510.21272 , archivePrefix =

    Pith/arXiv arXiv

  61. [61]

    Explain First, Trust Later:

    Watson, Adriana and Richards, Grant and Schiff, Daniel , journal=. Explain First, Trust Later:

  62. [62]

    2025 , eprint =

    Heterogeneous Graph Pre-training Based Model for Secure and Efficient Prediction of Default Risk Propagation among Bond Issuers , author =. 2025 , eprint =

    2025

  63. [63]

    Advances in Neural Information Processing Systems (NeurIPS) , year=

    Open Graph Benchmark: Datasets for Machine Learning on Graphs , author=. Advances in Neural Information Processing Systems (NeurIPS) , year=. 2005.00687 , archivePrefix=

    arXiv 2005

  64. [64]

    Advances in Neural Information Processing Systems (NeurIPS) , year=

    Temporal Graph Benchmark for Machine Learning on Temporal Graphs , author=. Advances in Neural Information Processing Systems (NeurIPS) , year=. 2307.01026 , archivePrefix=

    arXiv

  65. [65]

    arXiv preprint arXiv:2211.09110 , year=

    Holistic Evaluation of Language Models , author=. arXiv preprint arXiv:2211.09110 , year=. 2211.09110 , archivePrefix=

    Pith/arXiv arXiv

  66. [66]

    International Conference on Learning Representations (ICLR) , year=

    SWE-bench: Can Language Models Resolve Real-World GitHub Issues? , author=. International Conference on Learning Representations (ICLR) , year=. 2310.06770 , archivePrefix=

    Pith/arXiv arXiv

  67. [67]

    International Conference on Learning Representations (ICLR) , year=

    AgentBench: Evaluating LLMs as Agents , author=. International Conference on Learning Representations (ICLR) , year=. 2308.03688 , archivePrefix=

    Pith/arXiv arXiv

  68. [68]

    Proceedings of the 13th International Conference on Web Search and Data Mining , pages=

    DySAT: Deep Neural Representation Learning on Dynamic Graphs via Self-Attention Networks , author=. Proceedings of the 13th International Conference on Web Search and Data Mining , pages=. 2020 , doi=

    2020

  69. [69]

    International Conference on Learning Representations (ICLR) , year=

    Inductive Representation Learning on Temporal Graphs , author=. International Conference on Learning Representations (ICLR) , year=

  70. [70]

    2020 , doi=

    Pareja, Aldo and Domeniconi, Giacomo and Chen, Jie and Ma, Tengfei and Suzumura, Toyotaro and Kanezashi, Hiroki and Kaler, Tim and Schardl, Tao and Leiserson, Charles , booktitle=. 2020 , doi=

    2020

  71. [71]

    International Journal of Forecasting , volume=

    Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting , author=. International Journal of Forecasting , volume=. 2021 , doi=

    2021

  72. [72]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=. 2021 , doi=

    2021

  73. [73]

    Advances in Neural Information Processing Systems , volume=

    Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting , author=. Advances in Neural Information Processing Systems , volume=

  74. [74]

    International Conference on Learning Representations (ICLR) , year=

    Deep Graph Infomax , author=. International Conference on Learning Representations (ICLR) , year=

  75. [75]

    Advances in Neural Information Processing Systems , volume=

    Graph Contrastive Learning with Augmentations , author=. Advances in Neural Information Processing Systems , volume=

  76. [76]

    2022 , doi=

    Hou, Zhenyu and Liu, Xiao and Cen, Yukuo and Dong, Yuxiao and Yang, Hongxia and Wang, Chunjie and Tang, Jie , booktitle=. 2022 , doi=

    2022

  77. [77]

    2023 , url=

    Yao, Shunyu and Zhao, Jeffrey and Yu, Dian and Du, Nan and Shafran, Izhak and Narasimhan, Karthik and Cao, Yuan , booktitle=. 2023 , url=

    2023

  78. [78]

    2306.06031 , archivePrefix=

    Yang, Hongyang and Liu, Xiao-Yang and Wang, Christina Dan , year=. 2306.06031 , archivePrefix=

    arXiv

  79. [79]

    Auditing Asset-Specific Preferences in Financial Large Language Models: Evidence from

    Wu, Wenbin , year=. Auditing Asset-Specific Preferences in Financial Large Language Models: Evidence from. 2606.02528 , archivePrefix=

    Pith/arXiv arXiv