AI Paper Atlas

Research map / Time series / Causal and generative time series

KineFlow: Kinematic Second-Order Flow Matching for Time-Series Forecasting

Haiqi Jiang, Hui Xiong

ICML 2026 · Time series › Causal and generative time series · approach: Probabilistic and conformal forecasting

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add informative phase-space prior initialization instead of Gaussian noise (ablation)KineFlow · ablation avgMSE: 0.346 -> 0.302Improved
add informative phase-space prior initialization instead of Gaussian noise (ablation)KineFlow · ablation avgV-Err: 0.232 -> 0.166Improved
add velocity (second-order dynamics) to the state (ablation)KineFlow · ablation avgMSE: 0.333 -> 0.302Improved
add velocity (second-order dynamics) to the state (ablation)KineFlow · ablation avgCRPS: 0.337 -> 0.320Improved
add velocity (second-order dynamics) to the state (ablation)KineFlow · ablation avgV-Err: 0.214 -> 0.166Improved
initialize V0 from historical momentum instead of zero (ablation)KineFlow · ablation avgMSE: 0.311 -> 0.302Improved
initialize V0 from historical momentum instead of zero (ablation)KineFlow · ablation avgV-Err: 0.185 -> 0.166Improved
use KineFlow instead of TimeXerTimeXer · standard benchmarks avgMSE: 0.315 -> 0.269Improved
use KineFlow instead of TSFlowTSFlow · standard benchmarks avgCRPS: 0.332 -> 0.293Improved
use KineFlow instead of TimeXerTimeXer · AEMOMSE: 0.628 -> 0.400Improved
use KineFlow (0.90M params, few-shot) instead of Sundial zero-shotSundial · AEMO VIC to SAMSE: 0.285 -> 0.246Improved
use KineFlow instead of Electricity-ETTm1 best baseline on ETTm1 MSETimeXer · ETTm1MSE: 0.382 -> 0.397Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 5 of 5 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy