Towards Ideal Temporal Graph Neural Networks: Evaluations and Conclusions after 10,000 GPU Hours
Not on the map. Why?
Abstract (arXiv)
Temporal Graph Neural Networks (TGNNs) have emerged as powerful tools for modeling dynamic interactions across various domains. The design space of TGNNs is notably complex, given the unique challenges in runtime efficiency and scalability raised by the evolving nature of temporal graphs. We contend that many of the existing works on TGNN modeling inadequately explore the design space, leading to suboptimal designs. Viewing TGNN models through a performance-focused lens often obstructs a deeper understanding of the advantages and disadvantages of each technique. Specifically, benchmarking efforts inherently evaluate models in their original designs and implementations, resulting in unclear accuracy comparisons and misleading runtime. To address these shortcomings, we propose a practical comparative evaluation framework that performs a design space search across well-known TGNN modules based on a unified, optimized code implementation. Using our framework, we make the first efforts towards addressing three critical questions in TGNN design, spending over 10,000 GPU hours: (1) investigating the efficiency of TGNN module designs, (2) analyzing how the effectiveness of these modules correlates with dataset patterns, and (3) exploring the interplay between multiple modules. Key outcomes of this directed investigative approach include demonstrating that the most recent neighbor sampling and attention aggregator outperform uniform neighbor sampling and MLP-Mixer aggregator; Assessing static node memory as an effective node memory alternative, and showing that the choice between static or dynamic node memory should be based on the repetition patterns in the dataset. Our in-depth analysis of the interplay between TGNN modules and dataset patterns should provide a deeper insight into TGNN performance along with potential research directions for designing more general and effective TGNNs.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 32 of its 59 references that are in the map. First posted on arXiv 28 Dec 2024.
Builds on
32 of 59 references.
References · 32
- 2017Attention Is All You Needcited by 60,361
- 2017Inductive Representation Learning on Large Graphscited by 6,320
- 2021MLP-Mixer: An all-MLP Architecture for Visioncited by 1,446
- 2020Temporal Graph Networks for Deep Learning on Dynamic Graphscited by 530
- 2020Inductive Representation Learning on Temporal Graphscited by 462
- 2019Predicting Dynamic Embedding Trajectory in Temporal Interaction Networkscited by 448
- 2019EvolveGCN: Evolving Graph Convolutional Networks for Dynamic Graphscited by 447
- 2021Spectral Temporal Graph Neural Network for Multivariate Time-series Forecastingcited by 299
- 2019Representation Learning for Dynamic Graphs: A Surveycited by 265
- 2019Time2Vec: Learning a Vector Representation of Timecited by 213
- 2020Spatial-Temporal Fusion Graph Neural Networks for Traffic Flow Forecastingcited by 213
- 2021Inductive Representation Learning in Temporal Networks via Causal Anonymous Walkscited by 193
20 more
- 2022ROLAND: Graph Learning Framework for Dynamic Graphscited by 132
- 2022Towards Better Evaluation for Dynamic Link Predictioncited by 122
- 2021Sequential Recommendation with Graph Neural Networkscited by 89
- 2020APAN: Asynchronous Propagation Attention Network for Real-time Temporal Graph Embeddingcited by 78
- 2021Discrete-time Temporal Network Embedding via Implicit Hierarchical Learning in Hyperbolic Spacecited by 67
- 2022TGL: A General Framework for Temporal GNN Training on Billion-Scale Graphscited by 61
- 2020Structural Temporal Graph Neural Networks for Anomaly Detection in Dynamic Graphscited by 56
- 2022Algorithm and System Co-design for Efficient Subgraph-based Graph Representation Learningcited by 28
- 2023Deep learning for dynamic graphs: models and benchmarkscited by 19
- 2020Interaction-Based Trajectory Prediction Over a Hybrid Traffic Graphcited by 17
- 2020C-SAW: A Framework for Graph Sampling and Random Walk on GPUscited by 13
- 2023DistTGL: Distributed Memory-Based Temporal Graph Neural Network Trainingcited by 12
- 2021Efficient Dynamic Graph Representation Learning at Scalecited by 8
- 2023CAT-Walk: Inductive Hypergraph Learning via Set Walkscited by 6
- 2022GraphAD: A Graph Neural Network for Entity-Wise Multivariate Time-Series Anomaly Detectioncited by 5
- 2022DyG2Vec: Efficient Representation Learning for Dynamic Graphscited by 5
- 2023GNNFlow: A Distributed Framework for Continuous Temporal GNN Learning on Dynamic Graphscited by 5
- 2024On the Generalization Capability of Temporal Graph Learning Algorithms: Theoretical Insights and a Simpler Methodcited by 3
- 2022Bottleneck Analysis of Dynamic Graph Neural Network Inference on CPU and GPUcited by 2
- 2014User Session Identification Based on Strong Regularities in Inter-activity Timecited by 1
Built on by
Citing papers per year: 2025: 1; 2026: 1.
2 papers.