AI Paper Atlas

Research map / Time series / Causal and generative time series

KITE: Knowledge-Guided Probabilistic Modeling for Time Series Forecasting with Exogenous Variables

Hanyin Cheng, Jingrong Zhou, Yang Shu, Chenjuan Guo

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 History-Conditional Manifold (HCM) (ablation)KITE flow matching baseline · NP deterministicMSE: 0.395 -> 0.371Improved
add Knowledge-Guided Conditioning (KGC) (ablation)KITE flow matching baseline · NP deterministicMSE: 0.395 -> 0.384Improved
add KGC on top of HCM (ablation)KITE · NP deterministicMSE: 0.371 -> 0.330Improved
add Classifier-Free Guidance to HCM+KGC (ablation)KITE · NP deterministicMSE: 0.330 -> 0.325Improved
add HCM and KGC (ablation)KITE · Weather probabilisticCRPS: 0.087 -> 0.067Improved
add CFG to KGC without HCM (ablation)KITE · Weather probabilisticCRPS: 0.069 -> 0.065Improved
use Granger causality instead of Pearson for both exogenous types (ablation)KITE · NPMSE: 0.325 -> 0.334Hurt
use Granger for historical and Pearson for future exogenous (ablation)KITE · NPMSE: 0.325 -> 0.323Improved
use Pearson for historical and Granger for future exogenous (ablation)KITE · NPMSE: 0.325 -> 0.362Hurt
use KITE instead of K2VAEprobabilistic forecasting · eight datasetsCRPS: 5.7% reductionImproved
use KITE instead of CSDIprobabilistic forecasting · eight datasetsCRPS: 11.4% reductionImproved
use KITE instead of CrossLinear with future exogenousdeterministic forecasting · 12 datasetsMSE: 16.1% reductionImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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