Lune

ICML2025Top-tier venue

Arrow: Accelerator for Time Series Causal Discovery with Time Weaving

Yuanyuan Yao, Yuan Dong, Lu Chen, Kun Kuang, Ziquan Fang, Cheng Long, Yunjun Gao, Tianyi Li

Abstract

Current causal discovery methods for time series data can effectively address a variety of scenarios; however, they remain constrained by inefficiencies. The significant inefficiencies arise primarily from the high computational costs associated with binning, the uncertainty in selecting appropriate time lags, and the extensive sets of candidate variables. To achieve both high efficiency and effectiveness of causal discovery, we introduce an accelerator termed ARROW. It incorporates an innovative concept termed "Time Weaving" that efficiently encodes time series data to well capture the dynamic trends, thereby mitigating computational complexity. We also propose a novel time lag discovery strategy utilizing XOR operations, which derives a theorem to obtain the optimal time lag and significantly enhances the efficiency using XOR operations. To optimize the search space for causal relationships, we design an efficient pruning strategy that intelligently identifies the most relevant candidate variables, enhancing the efficiency and accuracy of causal discovery. We applied ARROW to four different types of time series causal discovery algorithms and evaluated it on 25 synthetic and real-world datasets. The results demonstrate that, compared to the original algorithms, ARROW achieves up to 153x speedup while achieving higher accuracy in most cases.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

2 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • CUTS+: High-Dimensional Causal Discovery from Irregular Time-Series cites High-recall causal discovery for autocorrelated time series with latent confounders
  • TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods cites Multiple Time Series Forecasting with Dynamic Graph Modeling
  • Enhancing Time Series Forecasting through Selective Representation Spaces: A Patch Perspective cites TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods
  • Coarse-to-Fine Learning of Dynamic Causal Structures cites High-recall causal discovery for autocorrelated time series with latent confounders
  • Coarse-to-Fine Learning of Dynamic Causal Structures cites CUTS+: High-Dimensional Causal Discovery from Irregular Time-Series
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers2

  1. Enhancing Time Series Forecasting through Selective Representation Spaces: A Patch PerspectiveXingjian Wu, Xiangfei Qiu, Hanyin Cheng, Zhengyu Li et al.NeurIPS 2025 · 58 citations
  2. Coarse-to-Fine Learning of Dynamic Causal StructuresDezhi Yang, Qiaoyu Tan, Carlotta Domeniconi, Jun Wang et al.ICLR 2026 · 2 citations

Builds on11

  1. TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting MethodsXiangfei Qiu, Jilin Hu, Lekui Zhou, Xingjian Wu et al.VLDB 2024 · 292 citations
  2. High-recall causal discovery for autocorrelated time series with latent confoundersAndreas Gerhardus, Jakob RungeNeurIPS 2020 · 159 citations
  3. Interpretable Models for Granger Causality Using Self-explaining Neural NetworksRicards Marcinkevics, Julia E. VogtICLR 2021 · 84 citations
  4. Multiple Time Series Forecasting with Dynamic Graph ModelingKai Zhao, Chenjuan Guo, Yunyao Cheng, Peng Han et al.VLDB 2024 · 67 citations
  5. CUTS+: High-Dimensional Causal Discovery from Irregular Time-SeriesYuxiao Cheng, Lianglong Li, Tingxiong Xiao, Zongren Li et al.AAAI 2024 · 58 citations

Related papers

  1. Influence without Confounding: Causal Discovery from Temporal Data with Long-term Carry-over EffectsFan Li, Zixuan Liu, Yi Zhao, Qi Tan et al.ICLR 2026
  2. GRACE-C: Generalized Rate Agnostic Causal Estimation via ConstraintsMohammadsajad Abavisani, David Danks, Sergey M. PlisICLR 2023
  3. Sparse Additive Model Pruning for Order-Based Causal Structure LearningKentaro Kanamori, Hirofumi Suzuki, Takuya TakagiAAAI 2026
  4. Structure Learning from Time-Series Data with Lag-Agnostic Structural PriorTaiyu Ban, Changxin Rong, Xiangyu Wang, Lyuzhou Chen et al.ICLR 2026
  5. Rhino: Deep Causal Temporal Relationship Learning with History-dependent NoiseWenbo Gong, Joel Jennings, Cheng Zhang, Nick PawlowskiICLR 2023 · 3 citations
Dusk over the sea between two cliffs drawn in fine vertical lines