Inductive Representation Learning on Temporal Graphs
read the original abstract
Inductive representation learning on temporal graphs is an important step toward salable machine learning on real-world dynamic networks. The evolving nature of temporal dynamic graphs requires handling new nodes as well as capturing temporal patterns. The node embeddings, which are now functions of time, should represent both the static node features and the evolving topological structures. Moreover, node and topological features can be temporal as well, whose patterns the node embeddings should also capture. We propose the temporal graph attention (TGAT) layer to efficiently aggregate temporal-topological neighborhood features as well as to learn the time-feature interactions. For TGAT, we use the self-attention mechanism as building block and develop a novel functional time encoding technique based on the classical Bochner's theorem from harmonic analysis. By stacking TGAT layers, the network recognizes the node embeddings as functions of time and is able to inductively infer embeddings for both new and observed nodes as the graph evolves. The proposed approach handles both node classification and link prediction task, and can be naturally extended to include the temporal edge features. We evaluate our method with transductive and inductive tasks under temporal settings with two benchmark and one industrial dataset. Our TGAT model compares favorably to state-of-the-art baselines as well as the previous temporal graph embedding approaches.
This paper has not been read by Pith yet.
Forward citations
Cited by 17 Pith papers
-
DRIFT: A Benchmark for Task-Free Continual Graph Learning with Continuous Distribution Shifts
DRIFT is a benchmark modeling continual graph data streams as time-varying mixtures of latent task distributions via Gaussian parameterization, revealing substantial performance degradation in existing continual learn...
-
DRIFT: A Benchmark for Task-Free Continual Graph Learning with Continuous Distribution Shifts
DRIFT is a benchmark for task-free continual graph learning under continuous distribution shifts, demonstrating that standard methods degrade without task boundary information.
-
Tencent Advertising Algorithm Challenge 2025: All-Modality Generative Recommendation
Releases TencentGR-1M and TencentGR-10M datasets with baselines for all-modality generative recommendation in advertising, including weighted evaluation for conversions.
-
Graph Retention Networks for Dynamic Graphs
Graph Retention Networks extend retention to dynamic graphs to enable parallelizable training, O(1) inference, and chunkwise long-term training while delivering competitive performance with major efficiency gains.
-
Forget Less, Generalize More: Unifying Temporal and Structural Adaptation for Dynamic Graphs
DSRD unifies temporal and structural adaptation for dynamic graphs via a single recurrent retentive state with learnable time-sensitivity parameters in the decay kernels.
-
Learning Dynamic Graph Representations through Timespan View Contrasts
CLDG and CLDG++ learn node representations on dynamic graphs by contrasting timespan views under temporal translation invariance, with extensions for global context via diffusion and integration into anomaly detection.
-
Dynamic Link Prediction with Temporally Enhanced Signed Graph Neural Networks
Introduces HCIM framework with recency weighting, LSTM trajectories, and temporal attention to enhance static signed GNNs, showing statistically significant gains on Bitcoin, Reddit, and synthetic TSN datasets.
-
DRIFT: A Benchmark for Task-Free Continual Graph Learning with Continuous Distribution Shifts
DRIFT benchmark shows substantial performance degradation for continual graph learning methods under task-free continuous distribution shifts modeled via Gaussian mixtures.
-
Time-varying Interaction Graph ODE for Dynamic Graph Representation Learning
TI-ODE decomposes graph ODE evolution into time-dependent learnable interaction basis functions, outperforming prior methods on dynamic graph attribute prediction and showing greater robustness.
-
BiTA: Bidirectional Gated Recurrent Unit-Transformer Aggregator in a Temporal Graph Network Framework for Alert Prediction in Computer Networks
BiTA redesigns temporal aggregation in TGNs by jointly using bidirectional GRU for sequential dependencies and Transformer for long-range context to improve alert prediction accuracy on real network data.
-
Unfolded Laplacian Spectral Embedding: A Theoretically Grounded Approach to Dynamic Network Representation
ULSE extends spectral embedding using normalized Laplacians with proven cross-sectional and longitudinal stability plus a dynamic Cheeger inequality under dynamic stochastic block models.
-
Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
Geometric deep learning provides a unified mathematical framework based on grids, groups, graphs, geodesics, and gauges to explain and extend neural network architectures by incorporating physical regularities.
-
Temporal Hyperbolic Graph Representation Learning for Scale-Free Internet Routing and Delay Prediction
HERMIT combines hyperbolic TGNN with RTT-aware edges and random forest regression to deliver 6% RMSE gain over historical baselines on 2015-2024 internet data while improving link prediction over prior hyperbolic models.
-
CHRONOS: Temporally-Aware Multi-Agent Coordination for Evolving Data Marketplaces
CHRONOS is a three-layer system for evolving data marketplaces that applies neural-ODE temporal decay, changepoint-aware Shapley valuation, and EXP3-IX private coordination to achieve 0.937 recall, 2.74 qps, 161 ms la...
-
A2QTGN: Adaptive Amplitude Quantum-Integrated Temporal Graph Network for Dynamic Link Prediction
A2QTGN combines adaptive quantum amplitude encoding with a temporal graph network to improve dynamic link prediction, showing strong results on five benchmark datasets.
-
Attention-based graph neural networks: a survey
The survey groups attention-based GNNs into three stages—graph recurrent attention networks, graph attention networks, and graph transformers—while reviewing architectures and future directions.
-
AdvSynGNN: Structure-Adaptive Graph Neural Nets via Adversarial Synthesis and Self-Corrective Propagation
AdvSynGNN uses multi-resolution structural synthesis, contrastive objectives, an adaptive transformer, and an adversarial propagation engine with residual label correction to improve node-level predictions on challeng...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.