AI Paper Atlas

Research map / Time series / Long-term forecasting

GCGNet: Graph-Consistent Generative Network for Time Series Forecasting with Exogenous Variables

Zhengyu Li, Xiangfei Qiu, Yuhan Zhu, Xingjian Wu, Jilin Hu, Chenjuan Guo, Bin Yang

ICLR 2026 · Time series › Long-term forecasting · approach: Graph-based multivariate forecasting · 3 citations (OpenAlex)

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
replace VAE with MLP in Variational Generator (adding the VAE) (ablation)GCGNet · NP, PJM, DE, Energy avgMSE: 0.363 -> 0.323Improved
add L_align to Graph Structure Aligner (ablation)GCGNet · NP, PJM, DE, Energy avgMSE: 0.474 -> 0.323Improved
use Graph VAE instead of deterministic Graph Learner (ablation)GCGNet · NP, PJM, DE, Energy avgMSE: 0.449 -> 0.323Improved
add Graph Refiner module (ablation)GCGNet · NP, PJM, DE, Energy avgMSE: 0.605 -> 0.323Improved
use VAE-based generative network under zero-masked exogenous values (ablation)GCGNet · Energy (zero mask)MSE: 0.142 -> 0.070Improved
use VAE-based generative network under random-masked exogenous values (ablation)GCGNet · NP (random mask)MSE: 0.272 -> 0.208Improved
use joint temporal-channel graph model versus best baselines with future exogenous variablesGCGNet · 12 datasetsMSE: 18 first places in MSE, 20 in MAEImproved
use GCGNet when future exogenous variables are unavailableGCGNet · 12 datasetsMSE: 8 first places in MSE vs 1 for TimeXer; Sdwpfm2 0.825 vs 0.803Mixed
use GCGNet versus baselines on Sdwpfm1 average MSEGCGNet · Sdwpfm1MSE: 0.416 vs 0.417 (CrossLinear)No clear effect
use GCGNet at long horizon on BEGCGNet · BE (horizon 360)MSE: 0.524 vs 0.482 (TFT)Hurt

Later papers that compared with this method

1 later accepted paper reports GCGNet as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 4 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