AI Paper Atlas

Research map / Benchmarks / PEMS08

PEMS08 leaderboard

24 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on PEMS08. Best MAE: 6.59 (WARP, ICLR 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

PEMS08 · MAE

24 papers, each with its best reported number for its own method; lower is better.

#MAE ↓MethodPaperTopic
16.59WARPWeight-Space Linear Recurrent Neural Networks
ICLR 2026 · PEMS08 table
Causal and generative time series
212.05SSL-STMFormerSSL-STMFormer Self-Supervised Learning Spatio-Temporal Entanglement Transformer for Traffic Flow Prediction
AAAI 2025 · Table 2
Video segmentation and tracking
312.23ST-VLMST-VLM: A Spatial-to-Image Multimodal Spatial-Temporal Prediction Framework with Vision-Language Model
AAAI 2026 · Table 1
Multimodal LLMs
412.38FedHINT
AGCRN encoders
Inter-Client Dependency Recovery with Hidden Global Components for Federated Traffic Prediction
AAAI 2026 · Table 1
Personalized and heterogeneous FL
512.58SIMBADSIMBAD: Spatio-Temporal Traffic Forecasting Robust to Aperiodicity
NeurIPS 2026 · Table 2
Knowledge and temporal graphs
613.04MetaDGMeta Dynamic Graph for Traffic Flow Prediction
AAAI 2026 · Table 2
Knowledge and temporal graphs
713.21LagLLM
GPT-2
LagLLM: LLM-empowered lead–lag dependency learning for spatial-temporal time series forecasting
ICML 2026 · Table 2
Long-term forecasting
813.31STD-PLM
GPT-2
STD-PLM: Understanding Both Spatial and Temporal Properties of Spatial-Temporal Data with PLM
AAAI 2025 · Forecasting performance table
Video segmentation and tracking
913.43EADSAn Expressive and Self-Adaptive Dynamical System for Efficient Function Learning
ICML 2025 · Table 1
RNNs and normalization
1013.583PDFormerPDFormer: Propagation Delay-Aware Dynamic Long-Range Transformer for Traffic Flow Prediction
AAAI 2023 · Performance on Graph-based Dat
Knowledge and temporal graphs
14 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

PEMS08 · RMSE

13 papers, each with its best reported number for its own method; lower is better.

#RMSE ↓MethodPaperTopic
110.1WARPWeight-Space Linear Recurrent Neural Networks
ICLR 2026 · PEMS08 table
Causal and generative time series
219.79ST-VLMST-VLM: A Spatial-to-Image Multimodal Spatial-Temporal Prediction Framework with Vision-Language Model
AAAI 2026 · Table 1
Multimodal LLMs
320.15SSL-STMFormerSSL-STMFormer Self-Supervised Learning Spatio-Temporal Entanglement Transformer for Traffic Flow Prediction
AAAI 2025 · Table 2
Video segmentation and tracking
420.52FedHINT
AGCRN encoders
Inter-Client Dependency Recovery with Hidden Global Components for Federated Traffic Prediction
AAAI 2026 · Table 1
Personalized and heterogeneous FL
522.13SIMBADSIMBAD: Spatio-Temporal Traffic Forecasting Robust to Aperiodicity
NeurIPS 2026 · Table 2
Knowledge and temporal graphs
622.53MetaDGMeta Dynamic Graph for Traffic Flow Prediction
AAAI 2026 · Table 2
Knowledge and temporal graphs
723.05STPGNNSpatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting
AAAI 2024 · Table 2
Knowledge and temporal graphs
823.19STD-PLM
GPT-2
STD-PLM: Understanding Both Spatial and Temporal Properties of Spatial-Temporal Data with PLM
AAAI 2025 · Forecasting performance table
Video segmentation and tracking
923.33RAST
MLP backbone
A Retrieval Augmented Spatio-Temporal Framework for Traffic Prediction
AAAI 2026 · Table 1
Knowledge and temporal graphs
1023.505PDFormerPDFormer: Propagation Delay-Aware Dynamic Long-Range Transformer for Traffic Flow Prediction
AAAI 2023 · Performance on Graph-based Dat
Knowledge and temporal graphs
3 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.