论文总结【LLM，时间序列基础模型，协变量，不规则时序，非平稳时序，稳健时序预测，不确定性量化等】)
ICML 2026将在2026年7月6日—11日于韩国首尔Seoul, South Korea举行。本文总结了2026 ICML上有关时间序列预测time series forecasting相关论文。如有疏漏欢迎大家补充。注由于时间序列标题包含time series或time-series的论文高达125篇其中两篇可以算作时空除去还有123篇笔者将分为上中下3篇推文来总结本文主要涉及时间序列预测Time Seires Forecasting, TSF的论文。时间序列预测TopicLLM时间序列基础模型不规则时序非平稳时序稳健时间序列预测不确定性量化等。1. L-Drive: Beyond a Single Mapping—Latent Context Drives Time Series Forecasting2. ConFlux: Multivariate Time Series in Flux, One Unified Forecast in Confluence3. PESD-TSF: A Period-Aware and Explicit Structured Decomposition Framework for Long-Term Time Series Forecasting4. Parameter Decorrelation via Transition-Variance Alignment for Multivariate Time-series Forecasting5. It’s TIME: Towards the Next Generation of Time Series Forecasting Benchmarks6. DropoutTS: Sample-Adaptive Dropout for Robust Time Series Forecasting7. Time-series forecasting through the lens of dynamics8. Benchmarking Physics-Informed Time-Series Models for Operational Global Station Weather Forecasting9. Byte Pair Encoding for Efficient Time Series Forecasting10. TimeGuard: Channel-wise Pool Training for Backdoor Defense in Time Series Forecasting11. What if Tomorrow is the World Cup Final? Counterfactual Time Series Forecasting with Textual Conditions12. Beyond Point Predictions: Manifold Expansion and Dual Alignment for Robust Time Series Distillation13. Channel Adapter for Time Series Foundation Models in Zero-Shot Multivariate Forecasting14. Beyond Extrapolation: Knowledge Utilization Paradigm with Bidirectional Inspiration for Time Series Forecasting15. Invariant Representation Learning for Source-Free Time Series Forecasting with LLM-Centric Proxy Denoising16. FIPN: Forward Self-Organizing Interpretable Polynomial Networks for Time Series Forecasting17. Anti-Aliasing Matters: A Dynamic Network for Time Series Forecasting18. DSENet: A Novel Dual-Stream Enhancement Network for Multi-Scale Non-Stationary Time Series Forecasting19. Crisp: A Spectral-Based Interaction Strategy for Multivariate Time Series Forecasting20. Beyond Model Ranking: Predictability-Aligned Evaluation for Time Series Forecasting21. KITE: Knowledge-Guided Probabilistic Modeling for Time Series Forecasting with Exogenous Variables22. TsLLM: Augmenting LLMs for General Time Series Understanding and Prediction23. ReNF: Rethinking the Principles of Neural Long-Term Time Series Forecasters24. Reviving Error Correction in Modern Deep Time-Series Forecasting25. MemCast: Memory-Driven Time Series Forecasting with Experience-Conditioned Reasoning26. Delving into Non-Exchangeability for Conformal Prediction in Graph-Structured Multivariate Time Series27. TSFAdv: Frequency-Guided Black-Box Adversarial Attacks on Time Series Forecasting28. Parametric Prior Mapping Framework for Non-stationary Probabilistic Time Series Forecasting29. Dynamic TMoE: A Drift-Aware Dynamic Mixture of Experts Framework for Non-Stationary Time Series Forecasting30. Ellipsoidal Time Series Forecasting31. Conditional Quantile Adjusted Conformal Prediction for Time Series32. CoGenCast: A Coupled Autoregressive–Flow Generative Framework for Time Series Forecasting33. StretchTime: Adaptive Time Series Forecasting via Symplectic Attention34. Robust Inter-Series Dependency Modeling for Time Series Forecasting via Information-Theoretic Alignment35. CombinationTS: A Modular Framework for Understanding Time-Series Forecasting Models36. See More, Forecast Better and Faster: Enhancing Time Series Foundation Models via Inference-Time Plug-and-Play Downsampling37. Taming the Recent-Data Bias: Towards Robust Time Series Forecasting with Global Context38. KineFlow: Kinematic Second-Order Flow Matching for Time-Series Forecasting39. From Observations to States: Latent Time Series Forecasting40. Baguan-TS: dual in-context learning model for time series forecasting with covariates41. Position: Current Benchmarking Hinders Real Progress in Deep Learning for Time Series Forecasting42. TimeMRA: LLM-Empowered Time Series Forecasting via Multi-Scale Retrieval-Augmented Representations43. Rethinking Multimodal Time-Series Forecasting Evaluation44. Bridging Time and Frequency: A Joint Modeling Framework for Irregular Multivariate Time Series Forecasting45. DAG: A Dual Correlation Network for Time Series Forecasting with Exogenous Variables46. SEER: Transformer-based Robust Time Series Forecasting via Automated Patch Enhancement and Replacement47. TimeSeed: Effective Time Series Forecasting with Sparse Endogenous Variables48. Information Geometry Loss for Time Series Forecasting49. KUMA: A Novel Framework with Koopman Separation and Efficient Multilevel Extraction in Time Series Forecasting50. Revealing Scaling Behavior in Large-scale Time Series Models: Implications for More Efficient and Accurate Forecasting51. PULSE: Generative Phase Evolution for Non-Stationary Time Series Forecasting52. Not All Frequencies Are Equal: Energy-Adaptive Diffusion for Time Series Forecasting53. MoRGEN: Mixture-of-Resolutions Generative Forecasting for Irregularly Sampled Medical Time-Series Data1 L-Drive: Beyond a Single Mapping—Latent Context Drives Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64484作者Fan Zhang ⋅ Shijun Chen ⋅ Hua Wang关键词隐式上下文驱动变化感知2 ConFlux: Multivariate Time Series in Flux, One Unified Forecast in Confluence链接https://icml.cc/virtual/2026/poster/61370作者Shiyu Wang ⋅ Yuchen Fang ⋅ Juntong Ni ⋅ Ziyi Zhang ⋅ Baichuan Mo ⋅ Xinyue Zhong ⋅ Chengxin Wang ⋅ Zhou Ye ⋅ Yang Xiang关键词百川归海3 PESD-TSF: A Period-Aware and Explicit Structured Decomposition Framework for Long-Term Time Series Forecasting链接https://icml.cc/virtual/2026/poster/65742作者Hua Wang ⋅ Xianhao jiao ⋅ Fan Zhang关键词周期感知显式结构分解4 Parameter Decorrelation via Transition-Variance Alignment for Multivariate Time-series Forecasting链接https://icml.cc/virtual/2026/poster/64647作者Ji-Eun Choi ⋅ Jae-Hong Lee ⋅ Joon Hyuk Chang关键词转移方差对齐马尔科夫链5 It’s TIME: Towards the Next Generation of Time Series Forecasting Benchmarks链接https://icml.cc/virtual/2026/poster/66086arXivhttp://arxiv.org/abs/2602.12147v3代码https://github.com/zqiao11/TIME作者Zhongzheng Qiao ⋅ SHENG PAN ⋅ Anni Wang ⋅ Viktoriya Zhukova ⋅ Yong Liu ⋅ Xudong Jiang ⋅ Qingsong Wen ⋅ Mingsheng Long ⋅ Ming Jin ⋅ Chenghao Liu关键词benchmark基础模型6 DropoutTS: Sample-Adaptive Dropout for Robust Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66019arXivhttp://arxiv.org/abs/2601.21726v1作者Siru Zhong ⋅ Yiqiu Liu ⋅ Zhiqing Cui ⋅ Zezhi Shao ⋅ Fei Wang ⋅ Qingsong Wen ⋅ Yuxuan Liang关键词频域样本自适应7 Time-series forecasting through the lens of dynamics链接https://icml.cc/virtual/2026/poster/64363arXivhttp://arxiv.org/abs/2507.15774v2作者Alexis-Raja Brachet ⋅ Pierre-Yves Richard ⋅ Céline Hudelot关键词数据底层动态8 Benchmarking Physics-Informed Time-Series Models for Operational Global Station Weather Forecasting链接https://icml.cc/virtual/2026/poster/65039arXivhttps://arxiv.org/abs/2406.14399代码https://github.com/taohan10200/WEATHER-5K作者Tao Han ⋅ Zhibin Wen ⋅ Zhenghao Chen ⋅ Dazhao Du ⋅ Song Guo ⋅ LEI BAI关键词天气预报物理驱动的时序模型9 Byte Pair Encoding for Efficient Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64868arXivhttp://arxiv.org/abs/2505.14411v3作者Leon Götz ⋅ Marcel Kollovieh ⋅ Stephan Günnemann ⋅ Leo Schwinn关键词token化解码10 TimeGuard: Channel-wise Pool Training for Backdoor Defense in Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64726作者Quang Duc Nguyen ⋅ Siyuan Liang ⋅ Yiming Li ⋅ Fushuo Huo ⋅ Dacheng Tao关键词后门防御11 What if Tomorrow is the World Cup Final? Counterfactual Time Series Forecasting with Textual Conditions链接https://icml.cc/virtual/2026/poster/65690作者Shuqi Gu ⋅ Yongxiang Zhao ⋅ Baoyu Jing ⋅ Kan Ren关键词反事实多模态12 Beyond Point Predictions: Manifold Expansion and Dual Alignment for Robust Time Series Distillation链接https://icml.cc/virtual/2026/poster/65540作者Junyao Hong ⋅ Zesheng Lai ⋅ Xinyi Xiao ⋅ Suyang Zhou ⋅ Aodong Shen ⋅ Youyong Kong关键词知识蒸馏双重对齐流形扩展13 Channel Adapter for Time Series Foundation Models in Zero-Shot Multivariate Forecasting链接https://icml.cc/virtual/2026/poster/64366作者Dongyuan Li ⋅ Renhe Jiang ⋅ Shun Zheng ⋅ Zheng Dong ⋅ Haotian Gao ⋅ Ying Zhang ⋅ Jiang Bian关键词通道自适应零样本14 Beyond Extrapolation: Knowledge Utilization Paradigm with Bidirectional Inspiration for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/63442作者Liu Chong ⋅ Yingjie Zhou ⋅ Hao Li ⋅ Pengyang Wang ⋅ Qingsong Wen ⋅ Ce Zhu关键词知识利用双向启发15 Invariant Representation Learning for Source-Free Time Series Forecasting with LLM-Centric Proxy Denoising链接https://icml.cc/virtual/2026/poster/66175作者Kangjia Yan ⋅ Chenxi Liu ⋅ Hao Miao ⋅ Xinle Wu ⋅ Yan Zhao ⋅ Chenjuan Guo ⋅ Bin Yang关键词迁移学习LLM16 FIPN: Forward Self-Organizing Interpretable Polynomial Networks for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64118作者YiZhen Wang ⋅ Zheng Wang ⋅ EUN-HU KIM ⋅ Zunwei Fu关键词自组织可解释多项式17 Anti-Aliasing Matters: A Dynamic Network for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/62513作者Heng Zhou ⋅ Xin Sun ⋅ Chao Li关键词频谱混叠动态网络18 DSENet: A Novel Dual-Stream Enhancement Network for Multi-Scale Non-Stationary Time Series Forecasting链接https://icml.cc/virtual/2026/poster/62164作者Yuhan Wang ⋅ Yuanyuan Zou ⋅ Jie Cheng ⋅ Bin Dai ⋅ Jinhong Guo关键词多尺度非平稳医疗时序19 Crisp: A Spectral-Based Interaction Strategy for Multivariate Time Series Forecasting链接https://icml.cc/virtual/2026/poster/63011作者Binwu Wang ⋅ Gaoyun Lin ⋅ Jiaming Ma ⋅ Qihe Huang ⋅ Zhengyang Zhou ⋅ Xu Wang ⋅ Pengkun Wang ⋅ Yang Wang关键词谱先验20 Beyond Model Ranking: Predictability-Aligned Evaluation for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64886arXivhttp://arxiv.org/abs/2509.23074v2作者Wanjin Feng ⋅ Yuan Yuan ⋅ Ding ⋅ Yong Li关键词线性利用率可预测漂移21 KITE: Knowledge-Guided Probabilistic Modeling for Time Series Forecasting with Exogenous Variables链接https://icml.cc/virtual/2026/poster/65231作者Hanyin Cheng ⋅ Jingrong Zhou ⋅ Yang Shu ⋅ Chenjuan Guo关键词协变量预测不确定性量化22 TsLLM: Augmenting LLMs for General Time Series Understanding and Prediction链接https://icml.cc/virtual/2026/poster/61098作者Felix Parker ⋅ Nimeesha Chan ⋅ Chi Zhang ⋅ Kimia Ghobadi关键词LLM时序理解23 ReNF: Rethinking the Principles of Neural Long-Term Time Series Forecasters链接https://icml.cc/virtual/2026/poster/62639作者Yihang Lu ⋅ Xianwei Meng ⋅ Enhong Chen关键词神经预测模型自回归24 Reviving Error Correction in Modern Deep Time-Series Forecasting链接https://icml.cc/virtual/2026/poster/63134作者Minh Nguyen ⋅ Van Dai Do ⋅ Huu Nguyen ⋅ Dung Nguyen ⋅ Kien Do ⋅ Hung Le关键词误差校正机制25 MemCast: Memory-Driven Time Series Forecasting with Experience-Conditioned Reasoning链接https://icml.cc/virtual/2026/poster/64890arXivhttp://arxiv.org/abs/2602.03164v1代码https://github.com/Xiaoyu-Tao/MemCast-TS作者Xiaoyu Tao ⋅ Mingyue Cheng ⋅ Ze Guo ⋅ Shuo Yu ⋅ Yaguo Liu ⋅ Qi Liu ⋅ Shijin Wang关键词记忆机制推理26 Delving into Non-Exchangeability for Conformal Prediction in Graph-Structured Multivariate Time Series链接https://icml.cc/virtual/2026/poster/66198arXivhttp://arxiv.org/abs/2605.04957v1作者Ruichao Guo ⋅ Xingyao Han ⋅ Wenshui Luo ⋅ Zhe Liu ⋅ Chen Gong ⋅ Hesheng Wang关键词共形预测不确定性量化27 TSFAdv: Frequency-Guided Black-Box Adversarial Attacks on Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66215作者Qizhuo Han ⋅ Xiangrui Cai ⋅ Sihan Xu ⋅ Ying Zhang ⋅ Zheli Liu关键词黑盒对抗攻击28 Parametric Prior Mapping Framework for Non-stationary Probabilistic Time Series Forecasting链接https://icml.cc/virtual/2026/poster/61351作者Jinglin Li ⋅ Jun Tan ⋅ QI Fang ⋅ Ning Gui关键词非平稳时序概率预测29 Dynamic TMoE: A Drift-Aware Dynamic Mixture of Experts Framework for Non-Stationary Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64808作者Jiawen Zhu ⋅ Shuhan Liu ⋅ Di Weng ⋅ Yingcai Wu关键词非平稳时序MoE30 Ellipsoidal Time Series Forecasting链接https://icml.cc/virtual/2026/poster/65519arXivhttp://arxiv.org/abs/2505.17370v6作者Qilin Wang关键词谱结构有效预测时长EPT31 Conditional Quantile Adjusted Conformal Prediction for Time Series链接https://icml.cc/virtual/2026/poster/64983作者Cheng Yu ⋅ Zhoufan Zhu ⋅ Ke Zhu关键词条件分位数共形预测32 CoGenCast: A Coupled Autoregressive–Flow Generative Framework for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64629arXivhttps://arxiv.org/abs/2602.03564作者Yaguo Liu ⋅ Mingyue Cheng ⋅ Daoyu Wang ⋅ Xiaoyu Tao ⋅ Qi Liu关键词自回归流匹配33 StretchTime: Adaptive Time Series Forecasting via Symplectic Attention链接https://icml.cc/virtual/2026/poster/66150作者Yubin Kim ⋅ Viresh Pati ⋅ Jevon Twitty ⋅ Vinh Pham ⋅ Shihao Yang ⋅ Jiecheng Lu关键词辛注意力旋转位置编码34 Robust Inter-Series Dependency Modeling for Time Series Forecasting via Information-Theoretic Alignment链接https://icml.cc/virtual/2026/poster/63281作者Wuqing Yu ⋅ Weichen Guo ⋅ Jian Zhou ⋅ Shuyu Luo ⋅ Jiacai Zhang关键词图 Transformer 周期性35 CombinationTS: A Modular Framework for Understanding Time-Series Forecasting Models链接https://icml.cc/virtual/2026/poster/65509arXivhttp://arxiv.org/abs/2605.01231v1作者Xiaorui Wang ⋅ Fanda Fan ⋅ Chenxi Wang ⋅ Yuxuan Yang ⋅ Rui Tang ⋅ Kuoyu Gao ⋅ simiao pang ⋅ Yuanfeng Shang ⋅ Liu ⋅ Gao ⋅ Lei Wang ⋅ Jianfeng Zhan关键词模块化归因36 See More, Forecast Better and Faster: Enhancing Time Series Foundation Models via Inference-Time Plug-and-Play Downsampling链接https://icml.cc/virtual/2026/poster/64120作者Longlong Xu ⋅ Zeyan Li ⋅ Xiao He ⋅ Zhaoyang Yu ⋅ Dazhong Wen ⋅ Mingze Sun ⋅ Changhua Pei ⋅ Dan Pei关键词时序基础模型下采样插件37 Taming the Recent-Data Bias: Towards Robust Time Series Forecasting with Global Context链接https://icml.cc/virtual/2026/poster/62481作者Longlong Xu ⋅ Zeyan Li ⋅ Xiao He ⋅ Zhaoyang Yu ⋅ Changhua Pei ⋅ Zhe Xie ⋅ Zijun Dou ⋅ Tieying Zhang ⋅ Dan Pei关键词近期数据偏差全局上下文38 KineFlow: Kinematic Second-Order Flow Matching for Time-Series Forecasting链接https://icml.cc/virtual/2026/poster/66194作者Haiqi Jiang ⋅ Hui Xiong关键词流匹配生成式运动学39 From Observations to States: Latent Time Series Forecasting链接https://icml.cc/virtual/2026/poster/61261arXivhttp://arxiv.org/abs/2602.00297v1代码https://github.com/Muyiiiii/LatentTSF作者Jie Yang ⋅ Yifan Hu ⋅ Yuante Li ⋅ Kexin Zhang ⋅ Kaize Ding ⋅ Philip Yu关键词隐空间时间序列预测40 Baguan-TS: dual in-context learning model for time series forecasting with covariates链接https://icml.cc/virtual/2026/poster/60703arXivhttps://arxiv.org/abs/2603.17439作者Linxiao Yang ⋅ Xue Jiang ⋅ Gezheng Xu ⋅ Tian Zhou ⋅ Min Yang ⋅ Zhaoyang Zhu ⋅ Linyuan Geng ⋅ Zhipeng Zeng ⋅ Qiming Chen ⋅ Xinyue Gu ⋅ Rong Jin ⋅ Liang Sun关键词上下文学习协变量41 Position: Current Benchmarking Hinders Real Progress in Deep Learning for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/67106arXivhttps://arxiv.org/abs/2512.22702作者Valentina Moretti ⋅ Andrea Cini ⋅ Ivan Marisca ⋅ Cesare Alippi关键词辅助预测模型卡片42 TimeMRA: LLM-Empowered Time Series Forecasting via Multi-Scale Retrieval-Augmented Representations链接https://icml.cc/virtual/2026/poster/62393作者Zongjiang Shang ⋅ Chengxi Jin ⋅ Binqing Wu ⋅ Dongliang Cui ⋅ Yue Yu ⋅ Haobang Sun ⋅ Chuanlin Xu ⋅ Ling Chen关键词LLMRAG43 Rethinking Multimodal Time-Series Forecasting Evaluation链接https://icml.cc/virtual/2026/poster/63015作者Haoxin Liu ⋅ Yichen Zhou ⋅ Rajat Sen ⋅ B. Aditya Prakash ⋅ Abhimanyu Das关键词多模态时序预测评估44 Bridging Time and Frequency: A Joint Modeling Framework for Irregular Multivariate Time Series Forecasting链接https://icml.cc/virtual/2026/poster/65148arXivhttp://arxiv.org/abs/2602.00582v1作者Xiangfei Qiu ⋅ Kangjia Yan ⋅ Xvyuan Liu ⋅ Xingjian Wu ⋅ Jilin Hu关键词不规则时序时频联合建模45 DAG: A Dual Correlation Network for Time Series Forecasting with Exogenous Variables链接https://icml.cc/virtual/2026/poster/62103arXivhttp://arxiv.org/abs/2509.14933v2作者Xiangfei Qiu ⋅ Yuhan Zhu ⋅ Zhengyu Li ⋅ Xingjian Wu ⋅ Bin Yang ⋅ Jilin Hu关键词协变量未来可知/近似可知协变量46 SEER: Transformer-based Robust Time Series Forecasting via Automated Patch Enhancement and Replacement链接https://icml.cc/virtual/2026/poster/61675arXivhttp://arxiv.org/abs/2602.00589v1作者Xiangfei Qiu ⋅ Xvyuan Liu ⋅ Tianen Shen ⋅ Xingjian Wu ⋅ Hanyin Cheng ⋅ Bin Yang ⋅ Jilin Hu关键词稳健性Patch增强与替换47 TimeSeed: Effective Time Series Forecasting with Sparse Endogenous Variables链接https://icml.cc/virtual/2026/poster/61541作者Zhaowang Wu ⋅ Kaixin Deng ⋅ Hua Yan关键词稀疏内生预测48 Information Geometry Loss for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66595作者Jiayu Fang ⋅ Xuande Liu ⋅ Sangsha Fang ⋅ Zhen Tian ⋅ Hongwei Ma ⋅ Zhiqi Shao ⋅ Junbin Gao关键词信息几何损失不确定性量化49 KUMA: A Novel Framework with Koopman Separation and Efficient Multilevel Extraction in Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64743作者Sijie Xiong ⋅ Cheng Tang ⋅ Atsushi Shimada关键词U 型多层注意力库普曼50 Revealing Scaling Behavior in Large-scale Time Series Models: Implications for More Efficient and Accurate Forecasting链接https://icml.cc/virtual/2026/poster/63243作者Xin Qiu ⋅ Junlong Tong ⋅ Yirong Sun ⋅ Yunpu Ma ⋅ Anhao Zhao关键词少层主导效应LLM4TS时序基础模型51 PULSE: Generative Phase Evolution for Non-Stationary Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64838作者Yangyou Liu ⋅ Zezhi Shao ⋅ Xinyu Chen ⋅ Hu Chen ⋅ Fei Wang ⋅ Yuankai Wu关键词非平稳时序相位52 Not All Frequencies Are Equal: Energy-Adaptive Diffusion for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66318作者Zining Qin ⋅ Huiling qin ⋅ Chenhao Wang ⋅ Jianxiong Guo ⋅ Tian Wang ⋅ Weijia Jia关键词扩散模型小波能量模型53 MoRGEN: Mixture-of-Resolutions Generative Forecasting for Irregularly Sampled Medical Time-Series Data链接https://icml.cc/virtual/2026/poster/64999作者Nassim Oufattole ⋅ Matthew McDermott ⋅ Collin Stultz关键词多分辨率混合生成模型不规则时序
ICML 2026 | 时间序列预测(TSF)论文总结【LLM，时间序列基础模型，协变量，不规则时序，非平稳时序，稳健时序预测，不确定性量化等】
发布时间：2026/7/5 2:30:05
ICML 2026将在2026年7月6日—11日于韩国首尔Seoul, South Korea举行。本文总结了2026 ICML上有关时间序列预测time series forecasting相关论文。如有疏漏欢迎大家补充。注由于时间序列标题包含time series或time-series的论文高达125篇其中两篇可以算作时空除去还有123篇笔者将分为上中下3篇推文来总结本文主要涉及时间序列预测Time Seires Forecasting, TSF的论文。时间序列预测TopicLLM时间序列基础模型不规则时序非平稳时序稳健时间序列预测不确定性量化等。1. L-Drive: Beyond a Single Mapping—Latent Context Drives Time Series Forecasting2. ConFlux: Multivariate Time Series in Flux, One Unified Forecast in Confluence3. PESD-TSF: A Period-Aware and Explicit Structured Decomposition Framework for Long-Term Time Series Forecasting4. Parameter Decorrelation via Transition-Variance Alignment for Multivariate Time-series Forecasting5. It’s TIME: Towards the Next Generation of Time Series Forecasting Benchmarks6. DropoutTS: Sample-Adaptive Dropout for Robust Time Series Forecasting7. Time-series forecasting through the lens of dynamics8. Benchmarking Physics-Informed Time-Series Models for Operational Global Station Weather Forecasting9. Byte Pair Encoding for Efficient Time Series Forecasting10. TimeGuard: Channel-wise Pool Training for Backdoor Defense in Time Series Forecasting11. What if Tomorrow is the World Cup Final? Counterfactual Time Series Forecasting with Textual Conditions12. Beyond Point Predictions: Manifold Expansion and Dual Alignment for Robust Time Series Distillation13. Channel Adapter for Time Series Foundation Models in Zero-Shot Multivariate Forecasting14. Beyond Extrapolation: Knowledge Utilization Paradigm with Bidirectional Inspiration for Time Series Forecasting15. Invariant Representation Learning for Source-Free Time Series Forecasting with LLM-Centric Proxy Denoising16. FIPN: Forward Self-Organizing Interpretable Polynomial Networks for Time Series Forecasting17. Anti-Aliasing Matters: A Dynamic Network for Time Series Forecasting18. DSENet: A Novel Dual-Stream Enhancement Network for Multi-Scale Non-Stationary Time Series Forecasting19. Crisp: A Spectral-Based Interaction Strategy for Multivariate Time Series Forecasting20. Beyond Model Ranking: Predictability-Aligned Evaluation for Time Series Forecasting21. KITE: Knowledge-Guided Probabilistic Modeling for Time Series Forecasting with Exogenous Variables22. TsLLM: Augmenting LLMs for General Time Series Understanding and Prediction23. ReNF: Rethinking the Principles of Neural Long-Term Time Series Forecasters24. Reviving Error Correction in Modern Deep Time-Series Forecasting25. MemCast: Memory-Driven Time Series Forecasting with Experience-Conditioned Reasoning26. Delving into Non-Exchangeability for Conformal Prediction in Graph-Structured Multivariate Time Series27. TSFAdv: Frequency-Guided Black-Box Adversarial Attacks on Time Series Forecasting28. Parametric Prior Mapping Framework for Non-stationary Probabilistic Time Series Forecasting29. Dynamic TMoE: A Drift-Aware Dynamic Mixture of Experts Framework for Non-Stationary Time Series Forecasting30. Ellipsoidal Time Series Forecasting31. Conditional Quantile Adjusted Conformal Prediction for Time Series32. CoGenCast: A Coupled Autoregressive–Flow Generative Framework for Time Series Forecasting33. StretchTime: Adaptive Time Series Forecasting via Symplectic Attention34. Robust Inter-Series Dependency Modeling for Time Series Forecasting via Information-Theoretic Alignment35. CombinationTS: A Modular Framework for Understanding Time-Series Forecasting Models36. See More, Forecast Better and Faster: Enhancing Time Series Foundation Models via Inference-Time Plug-and-Play Downsampling37. Taming the Recent-Data Bias: Towards Robust Time Series Forecasting with Global Context38. KineFlow: Kinematic Second-Order Flow Matching for Time-Series Forecasting39. From Observations to States: Latent Time Series Forecasting40. Baguan-TS: dual in-context learning model for time series forecasting with covariates41. Position: Current Benchmarking Hinders Real Progress in Deep Learning for Time Series Forecasting42. TimeMRA: LLM-Empowered Time Series Forecasting via Multi-Scale Retrieval-Augmented Representations43. Rethinking Multimodal Time-Series Forecasting Evaluation44. Bridging Time and Frequency: A Joint Modeling Framework for Irregular Multivariate Time Series Forecasting45. DAG: A Dual Correlation Network for Time Series Forecasting with Exogenous Variables46. SEER: Transformer-based Robust Time Series Forecasting via Automated Patch Enhancement and Replacement47. TimeSeed: Effective Time Series Forecasting with Sparse Endogenous Variables48. Information Geometry Loss for Time Series Forecasting49. KUMA: A Novel Framework with Koopman Separation and Efficient Multilevel Extraction in Time Series Forecasting50. Revealing Scaling Behavior in Large-scale Time Series Models: Implications for More Efficient and Accurate Forecasting51. PULSE: Generative Phase Evolution for Non-Stationary Time Series Forecasting52. Not All Frequencies Are Equal: Energy-Adaptive Diffusion for Time Series Forecasting53. MoRGEN: Mixture-of-Resolutions Generative Forecasting for Irregularly Sampled Medical Time-Series Data1 L-Drive: Beyond a Single Mapping—Latent Context Drives Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64484作者Fan Zhang ⋅ Shijun Chen ⋅ Hua Wang关键词隐式上下文驱动变化感知2 ConFlux: Multivariate Time Series in Flux, One Unified Forecast in Confluence链接https://icml.cc/virtual/2026/poster/61370作者Shiyu Wang ⋅ Yuchen Fang ⋅ Juntong Ni ⋅ Ziyi Zhang ⋅ Baichuan Mo ⋅ Xinyue Zhong ⋅ Chengxin Wang ⋅ Zhou Ye ⋅ Yang Xiang关键词百川归海3 PESD-TSF: A Period-Aware and Explicit Structured Decomposition Framework for Long-Term Time Series Forecasting链接https://icml.cc/virtual/2026/poster/65742作者Hua Wang ⋅ Xianhao jiao ⋅ Fan Zhang关键词周期感知显式结构分解4 Parameter Decorrelation via Transition-Variance Alignment for Multivariate Time-series Forecasting链接https://icml.cc/virtual/2026/poster/64647作者Ji-Eun Choi ⋅ Jae-Hong Lee ⋅ Joon Hyuk Chang关键词转移方差对齐马尔科夫链5 It’s TIME: Towards the Next Generation of Time Series Forecasting Benchmarks链接https://icml.cc/virtual/2026/poster/66086arXivhttp://arxiv.org/abs/2602.12147v3代码https://github.com/zqiao11/TIME作者Zhongzheng Qiao ⋅ SHENG PAN ⋅ Anni Wang ⋅ Viktoriya Zhukova ⋅ Yong Liu ⋅ Xudong Jiang ⋅ Qingsong Wen ⋅ Mingsheng Long ⋅ Ming Jin ⋅ Chenghao Liu关键词benchmark基础模型6 DropoutTS: Sample-Adaptive Dropout for Robust Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66019arXivhttp://arxiv.org/abs/2601.21726v1作者Siru Zhong ⋅ Yiqiu Liu ⋅ Zhiqing Cui ⋅ Zezhi Shao ⋅ Fei Wang ⋅ Qingsong Wen ⋅ Yuxuan Liang关键词频域样本自适应7 Time-series forecasting through the lens of dynamics链接https://icml.cc/virtual/2026/poster/64363arXivhttp://arxiv.org/abs/2507.15774v2作者Alexis-Raja Brachet ⋅ Pierre-Yves Richard ⋅ Céline Hudelot关键词数据底层动态8 Benchmarking Physics-Informed Time-Series Models for Operational Global Station Weather Forecasting链接https://icml.cc/virtual/2026/poster/65039arXivhttps://arxiv.org/abs/2406.14399代码https://github.com/taohan10200/WEATHER-5K作者Tao Han ⋅ Zhibin Wen ⋅ Zhenghao Chen ⋅ Dazhao Du ⋅ Song Guo ⋅ LEI BAI关键词天气预报物理驱动的时序模型9 Byte Pair Encoding for Efficient Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64868arXivhttp://arxiv.org/abs/2505.14411v3作者Leon Götz ⋅ Marcel Kollovieh ⋅ Stephan Günnemann ⋅ Leo Schwinn关键词token化解码10 TimeGuard: Channel-wise Pool Training for Backdoor Defense in Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64726作者Quang Duc Nguyen ⋅ Siyuan Liang ⋅ Yiming Li ⋅ Fushuo Huo ⋅ Dacheng Tao关键词后门防御11 What if Tomorrow is the World Cup Final? Counterfactual Time Series Forecasting with Textual Conditions链接https://icml.cc/virtual/2026/poster/65690作者Shuqi Gu ⋅ Yongxiang Zhao ⋅ Baoyu Jing ⋅ Kan Ren关键词反事实多模态12 Beyond Point Predictions: Manifold Expansion and Dual Alignment for Robust Time Series Distillation链接https://icml.cc/virtual/2026/poster/65540作者Junyao Hong ⋅ Zesheng Lai ⋅ Xinyi Xiao ⋅ Suyang Zhou ⋅ Aodong Shen ⋅ Youyong Kong关键词知识蒸馏双重对齐流形扩展13 Channel Adapter for Time Series Foundation Models in Zero-Shot Multivariate Forecasting链接https://icml.cc/virtual/2026/poster/64366作者Dongyuan Li ⋅ Renhe Jiang ⋅ Shun Zheng ⋅ Zheng Dong ⋅ Haotian Gao ⋅ Ying Zhang ⋅ Jiang Bian关键词通道自适应零样本14 Beyond Extrapolation: Knowledge Utilization Paradigm with Bidirectional Inspiration for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/63442作者Liu Chong ⋅ Yingjie Zhou ⋅ Hao Li ⋅ Pengyang Wang ⋅ Qingsong Wen ⋅ Ce Zhu关键词知识利用双向启发15 Invariant Representation Learning for Source-Free Time Series Forecasting with LLM-Centric Proxy Denoising链接https://icml.cc/virtual/2026/poster/66175作者Kangjia Yan ⋅ Chenxi Liu ⋅ Hao Miao ⋅ Xinle Wu ⋅ Yan Zhao ⋅ Chenjuan Guo ⋅ Bin Yang关键词迁移学习LLM16 FIPN: Forward Self-Organizing Interpretable Polynomial Networks for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64118作者YiZhen Wang ⋅ Zheng Wang ⋅ EUN-HU KIM ⋅ Zunwei Fu关键词自组织可解释多项式17 Anti-Aliasing Matters: A Dynamic Network for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/62513作者Heng Zhou ⋅ Xin Sun ⋅ Chao Li关键词频谱混叠动态网络18 DSENet: A Novel Dual-Stream Enhancement Network for Multi-Scale Non-Stationary Time Series Forecasting链接https://icml.cc/virtual/2026/poster/62164作者Yuhan Wang ⋅ Yuanyuan Zou ⋅ Jie Cheng ⋅ Bin Dai ⋅ Jinhong Guo关键词多尺度非平稳医疗时序19 Crisp: A Spectral-Based Interaction Strategy for Multivariate Time Series Forecasting链接https://icml.cc/virtual/2026/poster/63011作者Binwu Wang ⋅ Gaoyun Lin ⋅ Jiaming Ma ⋅ Qihe Huang ⋅ Zhengyang Zhou ⋅ Xu Wang ⋅ Pengkun Wang ⋅ Yang Wang关键词谱先验20 Beyond Model Ranking: Predictability-Aligned Evaluation for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64886arXivhttp://arxiv.org/abs/2509.23074v2作者Wanjin Feng ⋅ Yuan Yuan ⋅ Ding ⋅ Yong Li关键词线性利用率可预测漂移21 KITE: Knowledge-Guided Probabilistic Modeling for Time Series Forecasting with Exogenous Variables链接https://icml.cc/virtual/2026/poster/65231作者Hanyin Cheng ⋅ Jingrong Zhou ⋅ Yang Shu ⋅ Chenjuan Guo关键词协变量预测不确定性量化22 TsLLM: Augmenting LLMs for General Time Series Understanding and Prediction链接https://icml.cc/virtual/2026/poster/61098作者Felix Parker ⋅ Nimeesha Chan ⋅ Chi Zhang ⋅ Kimia Ghobadi关键词LLM时序理解23 ReNF: Rethinking the Principles of Neural Long-Term Time Series Forecasters链接https://icml.cc/virtual/2026/poster/62639作者Yihang Lu ⋅ Xianwei Meng ⋅ Enhong Chen关键词神经预测模型自回归24 Reviving Error Correction in Modern Deep Time-Series Forecasting链接https://icml.cc/virtual/2026/poster/63134作者Minh Nguyen ⋅ Van Dai Do ⋅ Huu Nguyen ⋅ Dung Nguyen ⋅ Kien Do ⋅ Hung Le关键词误差校正机制25 MemCast: Memory-Driven Time Series Forecasting with Experience-Conditioned Reasoning链接https://icml.cc/virtual/2026/poster/64890arXivhttp://arxiv.org/abs/2602.03164v1代码https://github.com/Xiaoyu-Tao/MemCast-TS作者Xiaoyu Tao ⋅ Mingyue Cheng ⋅ Ze Guo ⋅ Shuo Yu ⋅ Yaguo Liu ⋅ Qi Liu ⋅ Shijin Wang关键词记忆机制推理26 Delving into Non-Exchangeability for Conformal Prediction in Graph-Structured Multivariate Time Series链接https://icml.cc/virtual/2026/poster/66198arXivhttp://arxiv.org/abs/2605.04957v1作者Ruichao Guo ⋅ Xingyao Han ⋅ Wenshui Luo ⋅ Zhe Liu ⋅ Chen Gong ⋅ Hesheng Wang关键词共形预测不确定性量化27 TSFAdv: Frequency-Guided Black-Box Adversarial Attacks on Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66215作者Qizhuo Han ⋅ Xiangrui Cai ⋅ Sihan Xu ⋅ Ying Zhang ⋅ Zheli Liu关键词黑盒对抗攻击28 Parametric Prior Mapping Framework for Non-stationary Probabilistic Time Series Forecasting链接https://icml.cc/virtual/2026/poster/61351作者Jinglin Li ⋅ Jun Tan ⋅ QI Fang ⋅ Ning Gui关键词非平稳时序概率预测29 Dynamic TMoE: A Drift-Aware Dynamic Mixture of Experts Framework for Non-Stationary Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64808作者Jiawen Zhu ⋅ Shuhan Liu ⋅ Di Weng ⋅ Yingcai Wu关键词非平稳时序MoE30 Ellipsoidal Time Series Forecasting链接https://icml.cc/virtual/2026/poster/65519arXivhttp://arxiv.org/abs/2505.17370v6作者Qilin Wang关键词谱结构有效预测时长EPT31 Conditional Quantile Adjusted Conformal Prediction for Time Series链接https://icml.cc/virtual/2026/poster/64983作者Cheng Yu ⋅ Zhoufan Zhu ⋅ Ke Zhu关键词条件分位数共形预测32 CoGenCast: A Coupled Autoregressive–Flow Generative Framework for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64629arXivhttps://arxiv.org/abs/2602.03564作者Yaguo Liu ⋅ Mingyue Cheng ⋅ Daoyu Wang ⋅ Xiaoyu Tao ⋅ Qi Liu关键词自回归流匹配33 StretchTime: Adaptive Time Series Forecasting via Symplectic Attention链接https://icml.cc/virtual/2026/poster/66150作者Yubin Kim ⋅ Viresh Pati ⋅ Jevon Twitty ⋅ Vinh Pham ⋅ Shihao Yang ⋅ Jiecheng Lu关键词辛注意力旋转位置编码34 Robust Inter-Series Dependency Modeling for Time Series Forecasting via Information-Theoretic Alignment链接https://icml.cc/virtual/2026/poster/63281作者Wuqing Yu ⋅ Weichen Guo ⋅ Jian Zhou ⋅ Shuyu Luo ⋅ Jiacai Zhang关键词图 Transformer 周期性35 CombinationTS: A Modular Framework for Understanding Time-Series Forecasting Models链接https://icml.cc/virtual/2026/poster/65509arXivhttp://arxiv.org/abs/2605.01231v1作者Xiaorui Wang ⋅ Fanda Fan ⋅ Chenxi Wang ⋅ Yuxuan Yang ⋅ Rui Tang ⋅ Kuoyu Gao ⋅ simiao pang ⋅ Yuanfeng Shang ⋅ Liu ⋅ Gao ⋅ Lei Wang ⋅ Jianfeng Zhan关键词模块化归因36 See More, Forecast Better and Faster: Enhancing Time Series Foundation Models via Inference-Time Plug-and-Play Downsampling链接https://icml.cc/virtual/2026/poster/64120作者Longlong Xu ⋅ Zeyan Li ⋅ Xiao He ⋅ Zhaoyang Yu ⋅ Dazhong Wen ⋅ Mingze Sun ⋅ Changhua Pei ⋅ Dan Pei关键词时序基础模型下采样插件37 Taming the Recent-Data Bias: Towards Robust Time Series Forecasting with Global Context链接https://icml.cc/virtual/2026/poster/62481作者Longlong Xu ⋅ Zeyan Li ⋅ Xiao He ⋅ Zhaoyang Yu ⋅ Changhua Pei ⋅ Zhe Xie ⋅ Zijun Dou ⋅ Tieying Zhang ⋅ Dan Pei关键词近期数据偏差全局上下文38 KineFlow: Kinematic Second-Order Flow Matching for Time-Series Forecasting链接https://icml.cc/virtual/2026/poster/66194作者Haiqi Jiang ⋅ Hui Xiong关键词流匹配生成式运动学39 From Observations to States: Latent Time Series Forecasting链接https://icml.cc/virtual/2026/poster/61261arXivhttp://arxiv.org/abs/2602.00297v1代码https://github.com/Muyiiiii/LatentTSF作者Jie Yang ⋅ Yifan Hu ⋅ Yuante Li ⋅ Kexin Zhang ⋅ Kaize Ding ⋅ Philip Yu关键词隐空间时间序列预测40 Baguan-TS: dual in-context learning model for time series forecasting with covariates链接https://icml.cc/virtual/2026/poster/60703arXivhttps://arxiv.org/abs/2603.17439作者Linxiao Yang ⋅ Xue Jiang ⋅ Gezheng Xu ⋅ Tian Zhou ⋅ Min Yang ⋅ Zhaoyang Zhu ⋅ Linyuan Geng ⋅ Zhipeng Zeng ⋅ Qiming Chen ⋅ Xinyue Gu ⋅ Rong Jin ⋅ Liang Sun关键词上下文学习协变量41 Position: Current Benchmarking Hinders Real Progress in Deep Learning for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/67106arXivhttps://arxiv.org/abs/2512.22702作者Valentina Moretti ⋅ Andrea Cini ⋅ Ivan Marisca ⋅ Cesare Alippi关键词辅助预测模型卡片42 TimeMRA: LLM-Empowered Time Series Forecasting via Multi-Scale Retrieval-Augmented Representations链接https://icml.cc/virtual/2026/poster/62393作者Zongjiang Shang ⋅ Chengxi Jin ⋅ Binqing Wu ⋅ Dongliang Cui ⋅ Yue Yu ⋅ Haobang Sun ⋅ Chuanlin Xu ⋅ Ling Chen关键词LLMRAG43 Rethinking Multimodal Time-Series Forecasting Evaluation链接https://icml.cc/virtual/2026/poster/63015作者Haoxin Liu ⋅ Yichen Zhou ⋅ Rajat Sen ⋅ B. Aditya Prakash ⋅ Abhimanyu Das关键词多模态时序预测评估44 Bridging Time and Frequency: A Joint Modeling Framework for Irregular Multivariate Time Series Forecasting链接https://icml.cc/virtual/2026/poster/65148arXivhttp://arxiv.org/abs/2602.00582v1作者Xiangfei Qiu ⋅ Kangjia Yan ⋅ Xvyuan Liu ⋅ Xingjian Wu ⋅ Jilin Hu关键词不规则时序时频联合建模45 DAG: A Dual Correlation Network for Time Series Forecasting with Exogenous Variables链接https://icml.cc/virtual/2026/poster/62103arXivhttp://arxiv.org/abs/2509.14933v2作者Xiangfei Qiu ⋅ Yuhan Zhu ⋅ Zhengyu Li ⋅ Xingjian Wu ⋅ Bin Yang ⋅ Jilin Hu关键词协变量未来可知/近似可知协变量46 SEER: Transformer-based Robust Time Series Forecasting via Automated Patch Enhancement and Replacement链接https://icml.cc/virtual/2026/poster/61675arXivhttp://arxiv.org/abs/2602.00589v1作者Xiangfei Qiu ⋅ Xvyuan Liu ⋅ Tianen Shen ⋅ Xingjian Wu ⋅ Hanyin Cheng ⋅ Bin Yang ⋅ Jilin Hu关键词稳健性Patch增强与替换47 TimeSeed: Effective Time Series Forecasting with Sparse Endogenous Variables链接https://icml.cc/virtual/2026/poster/61541作者Zhaowang Wu ⋅ Kaixin Deng ⋅ Hua Yan关键词稀疏内生预测48 Information Geometry Loss for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66595作者Jiayu Fang ⋅ Xuande Liu ⋅ Sangsha Fang ⋅ Zhen Tian ⋅ Hongwei Ma ⋅ Zhiqi Shao ⋅ Junbin Gao关键词信息几何损失不确定性量化49 KUMA: A Novel Framework with Koopman Separation and Efficient Multilevel Extraction in Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64743作者Sijie Xiong ⋅ Cheng Tang ⋅ Atsushi Shimada关键词U 型多层注意力库普曼50 Revealing Scaling Behavior in Large-scale Time Series Models: Implications for More Efficient and Accurate Forecasting链接https://icml.cc/virtual/2026/poster/63243作者Xin Qiu ⋅ Junlong Tong ⋅ Yirong Sun ⋅ Yunpu Ma ⋅ Anhao Zhao关键词少层主导效应LLM4TS时序基础模型51 PULSE: Generative Phase Evolution for Non-Stationary Time Series Forecasting链接https://icml.cc/virtual/2026/poster/64838作者Yangyou Liu ⋅ Zezhi Shao ⋅ Xinyu Chen ⋅ Hu Chen ⋅ Fei Wang ⋅ Yuankai Wu关键词非平稳时序相位52 Not All Frequencies Are Equal: Energy-Adaptive Diffusion for Time Series Forecasting链接https://icml.cc/virtual/2026/poster/66318作者Zining Qin ⋅ Huiling qin ⋅ Chenhao Wang ⋅ Jianxiong Guo ⋅ Tian Wang ⋅ Weijia Jia关键词扩散模型小波能量模型53 MoRGEN: Mixture-of-Resolutions Generative Forecasting for Irregularly Sampled Medical Time-Series Data链接https://icml.cc/virtual/2026/poster/64999作者Nassim Oufattole ⋅ Matthew McDermott ⋅ Collin Stultz关键词多分辨率混合生成模型不规则时序
