时间序列论文趋势 2023–2026

主要结论

• 时间序列论文从 2023 年的 272 篇增长到 2026 年的 876 篇，三年间扩大到约 3.2 倍。
• NeurIPS 是发时间序列论文最多的会议，2023–2026 年共 757 篇，占四个会议总量的 35%。
• 交叉方向里涨得最快的是多模态大模型：同时属于两个主题的论文从 2023 年的 6 篇增至 2026 年的 85 篇。
• 在「图与结构化数据」的 5 个主题里，它的 2026 年论文数排第 1；同组增长最快的分子建模是 2.7 倍，时间序列是 3.2 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026A Batched Hartigan's $k$-Means Clustering
• ICML 2026A Control-Theoretic View of Mamba on Stability and Robustness
• ICLR 2026A Single Architecture for Representing Invariance Under Any Space Group
• CVPR 2026Activation Matters Test-time Activated Negative Labels for OOD Detection with
• NeurIPS 2025Abstain Mask Retain Core: Time Series Prediction by Adaptive Masking Loss with Representation Consistency
• ICML 2025$K^2$VAE: A Koopman-Kalman Enhanced Variational AutoEncoder for Probabilistic Time Series Forecasting
• ICLR 2025A deep inverse-mapping model for a flapping robotic wing
• CVPR 2025Beyond Single-Modal Boundary: Cross-Modal Anomaly Detection through Visual Prototype and Harmonization
• NeurIPS 2024A Continuous-time Stochastic Gradient Descent Method for Continuous Data
• ICML 2024$S^2$IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting
• ICLR 2024ARM: Refining Multivariate Forecasting with Adaptive Temporal-Contextual Learning
• CVPR 2024Each Test Image Deserves A Specific Prompt: Continual Test-Time Adaptation for 2D Medical Image Segmentation
• NeurIPS 2023A Sublinear-Time Spectral Clustering Oracle with Improved Preprocessing Time
• ICML 2023Accounting For Informative Sampling When Learning to Forecast Treatment Outcomes Over Time
• ICLR 2023A Time Series is Worth 64 Words: Long-term Forecasting with Transformers
• CVPR 2023ActMAD: Activation Matching To Align Distributions for Test-Time-Training
• NeurIPS 2026A Locally Tokenized Generative Model for Robust Time-Series Watermarking
• ICML 2026ADHD Disease Detection Based on Short- and Long-Term Brain Function Encoding and Memory Graph Network
• ICLR 2026A Spectral-Grassmann Wasserstein metric for operator representations of dynamical systems
• CVPR 2026Continuous Exposure-Time Modeling for Realistic Atmospheric Turbulence Synthesis

全部时间序列论文

共 264 篇，第 6 / 6 页

251. ICML 2026TsLLM: Augmenting LLMs for General Time Series Understanding and Prediction
252. ICML 2026TSRBench: A Comprehensive Multi-task Multi-modal Time Series Reasoning Benchmark for Generalist Models
253. ICML 2026Unfolded Laplacian Spectral Embedding: A Theoretically Grounded Approach to Dynamic Network Representation
254. ICML 2026Unified Time Series Explanations via Amortized Optimization and Instance-level Multi-Expert Knowledge Distillation
255. ICML 2026Universal Redundancies in Time Series Foundation Models
256. ICML 2026WaveSSM: Multiscale State-Space Models for Non-stationary Signal Attention
257. ICML 2026What if Tomorrow is the World Cup Final? Counterfactual Time Series Forecasting with Textual Conditions
258. ICML 2026What If We Allocate Test-Time Compute Adaptively?
259. ICML 2026What If We Let Forecasting Forget? A Sparse Bottleneck for Cross-Variable Dependencies
260. ICML 2026WildCat: Near-Linear Attention in Theory and Practice
261. ICML 2026Winformer: Transcending Pairwise Similarity for Time-series Generation
262. ICML 2026Words Towards Explainability: Caption Label-Free Learning via Dual Loop Agentic Time Series Captioning
263. ICML 2026ZeroDiff: Zero-Shot Time Series Reconstruction via Informed-Prior Diffusion
264. ICML 2026Zeus: Towards Tuning-Free Foundation Model for Time Series Analysis

常见问题

2023–2026 年四大会议一共有多少篇时间序列论文？

共 2,157 篇，按年份分别是 2023 年 272 篇、2024 年 424 篇、2025 年 585 篇、2026 年 876 篇。

NeurIPS 2026 有多少篇时间序列论文？

NeurIPS 2026 有 298 篇时间序列论文，上一届是 221 篇。

ICLR 2026 有多少篇时间序列论文？

ICLR 2026 有 220 篇时间序列论文，上一届是 131 篇。

ICML 2026 有多少篇时间序列论文？

ICML 2026 有 264 篇时间序列论文，上一届是 159 篇。

CVPR 2026 有多少篇时间序列论文？

CVPR 2026 有 94 篇时间序列论文，上一届是 74 篇。

哪个会议的时间序列论文最多？

NeurIPS，2023–2026 年共 757 篇。