AI Paper Atlas

Research map / Probabilistic and scientific modeling / Dynamical systems

SONATA: Synergistic Coreset Informed Adaptive Temporal Tensor Factorization

Maolin Wang, Zhiqi Li, Binhao Wang, Xuhui Chen, Tianshuo Wei, Wanyu Wang, Shikai Fang, Ruocheng Guo, Zenglin Xu, Xiangyu Zhao

ICLR 2026 · Probabilistic and scientific modeling › Dynamical systems · approach: Latent dynamics models

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
use SONATA instead of second-best SFTL-CPstreaming tensor factorization · CA Traffic 30KRMSE: 0.231 -> 0.089Improved
use SONATA instead of NONFATtensor factorization · ServerRoomRMSE: 0.117 -> 0.115No clear effect
use Matern-3/2 kernel instead of Matern-1/2 (ablation)SONATA · ServerRoomRMSE: reducing RMSE by 49.1% and MAE by 47.0%Improved
choose lengthscale 0.3 instead of 0.1 or 0.9 (ablation)SONATA · ServerRoomRMSE: 0.1293 (0.3) vs 0.1830 (0.1), 0.1474 (0.9)Improved
use synergistic coreset selection instead of random sampling (ablation)SONATA · ServerRoom, 400 samplesRMSE: 0.1481 -> 0.1293Improved
use synergistic coreset instead of processing all data points (ablation)SONATA · ServerRoomRMSE: RMSE: 0.1290 vs 0.1293No clear effect
increase coreset budget Mmax from 1000 to 2000 (ablation)SONATA · CA TrafficRMSE: 0.1808 -> 0.0938Improved
increase coreset budget Mmax from 2000 to 3000 (ablation)SONATA · CA TrafficRMSE: 0.0938 -> 0.0891No clear effect
set discount factor to 0.5 instead of 0.9 (ablation)SONATA · ServerRoomRMSE: 0.1293 -> 0.1156Improved
set discount factor to 0.9 instead of 0.5 (ablation)SONATA · CA TrafficRMSE: 0.1107 -> 0.0893Improved
use nonzero discount factor instead of 0 (ablation)SONATA · CA TrafficRMSE: 0.1716 -> 0.0893Improved

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