mlsanityData to 14 Oct 2026

ST-ExpertNet: A Deep Expert Framework for Traffic Prediction

arXiv 2205.07851 ↗First posted 5 May 2022cs.LG cs.AIcited by 10

On the map: Mixture-of-experts → MoE for time series and graphs

Abstract (arXiv)

Recently, forecasting the crowd flows has become an important research topic, and plentiful technologies have achieved good performances. As we all know, the flow at a citywide level is in a mixed state with several basic patterns (e.g., commuting, working, and commercial) caused by the city area functional distributions (e.g., developed commercial areas, educational areas and parks). However, existing technologies have been criticized for their lack of considering the differences in the flow patterns among regions since they want to build only one comprehensive model to learn the mixed flow tensors. Recognizing this limitation, we present a new perspective on flow prediction and propose an explainable framework named ST-ExpertNet, which can adopt every spatial-temporal model and train a set of functional experts devoted to specific flow patterns. Technically, we train a bunch of experts based on the Mixture of Experts (MoE), which guides each expert to specialize in different kinds of flow patterns in sample spaces by using the gating network. We define several criteria, including comprehensiveness, sparsity, and preciseness, to construct the experts for better interpretability and performances. We conduct experiments on a wide range of real-world taxi and bike datasets in Beijing and NYC. The visualizations of the expert's intermediate results demonstrate that our ST-ExpertNet successfully disentangles the city's mixed flow tensors along with the city layout, e.g., the urban ring road structure. Different network architectures, such as ST-ResNet, ConvLSTM, and CNN, have been adopted into our ST-ExpertNet framework for experiments and the results demonstrates the superiority of our framework in both interpretability and performances.

Key facts. Cited by 10 papers in the map (data through 14 Oct 2026). Placed in MoE for time series and graphs, one of 12 directions in Mixture-of-experts. Builds on 14 of its 59 references that are in the map. First posted on arXiv 5 May 2022.

Builds on

14 of 59 references.

References · 14

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  3. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  4. 2014Neural Machine Translation by Jointly Learning to Align and Translatecited by 4,548
  5. 2015Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcastingcited by 1,436
  6. 2016Conditional Image Generation with PixelCNN Decoderscited by 862
  7. 2016Expert Gate: Lifelong Learning with a Network of Expertscited by 372
  8. 2016Deep Spatio-Temporal Residual Networks for Citywide Crowd Flows Predictioncited by 333
  9. 2019Improving Adversarial Robustness via Promoting Ensemble Diversitycited by 213
  10. 2018Deep Multi-View Spatial-Temporal Network for Taxi Demand Predictioncited by 162
  11. 2015Delving Deeper into Convolutional Networks for Learning Video Representationscited by 148
  12. 2018Revisiting Spatial-Temporal Similarity: A Deep Learning Framework for Traffic Predictioncited by 122
2 more
  1. 2021DL-Traff: Survey and Benchmark of Deep Learning Models for Urban Traffic Predictioncited by 73
  2. 2017Predicting Citywide Crowd Flows Using Deep Spatio-Temporal Residual Networkscited by 34

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 1; 2024: 5; 2025: 3; 2026: 1.

10 papers.

  1. 2025Mixture of Experts (MoE): A Big Data Perspectivecited by 18
  2. 2024Evaluating the Generalization Ability of Spatiotemporal Model in Urban Scenariocited by 7
  3. 2024Integration of Mixture of Experts and Multimodal Generative AI in Internet of Vehicles: A Surveycited by 6
  4. 2024Continual Traffic Forecasting via Mixture of Expertscited by 6
  5. 2023Causal-Based Supervision of Attention in Graph Neural Network: A Better and Simpler Choice towards Powerful Attentioncited by 5
  6. 2024DisenTS: Disentangled Channel Evolving Pattern Modeling for Multivariate Time Series Forecastingcited by 3
  7. 2024Understanding Token-level Topological Structures in Transformer-based Time Series Forecastingcited by 1
  8. 2026Ground-Truth Neighborhood Regularization for Reinforcement Learning Post-Training of Time Series Foundation Modelscited by 1
  9. 2025City-Level Foreign Direct Investment Prediction with Tabular Learning on Judicial Datacited by 0
  10. 2025Towards Resilient Transportation: A Conditional Transformer for Accident-Informed Traffic Forecastingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024A Time Series is Worth Five Experts: Heterogeneous Mixture of Experts for Traffic Flow Prediction
  2. 2025FlowDistill: Scalable Traffic Flow Prediction via Distillation from LLMs
  3. 2025Efficient Traffic Prediction Through Spatio-Temporal Distillation
  4. 2026STHMoE: Hypergraph-Enhanced Heterogeneous Dependency Coordination for LLM-Based Urban Traffic Data Forecasting
  5. 2025Less is More: Strategic Expert Selection Outperforms Ensemble Complexity in Traffic Forecasting
  6. 2026MoE Enhanced Federated Learning for Spatiotemporal Prediction
  7. 2024Navigating Spatio-Temporal Heterogeneity: A Graph Transformer Approach for Traffic Forecasting
  8. 2024ST-MambaSync: The Complement of Mamba and Transformers for Spatial-Temporal in Traffic Flow Prediction