Temporal Link Prediction

Follow this topic Ask about this topic

Temporal link prediction estimates future interactions or facts from evolving networks, helping forecast communication, recommendations, biological relations, and knowledge-graph events. Methods range from temporal graph neural networks with memory, attention, or state-space models to motif and structural features, temporal knowledge-graph embeddings, and event-process models. Recent work also tests lightweight alternatives such as smoothed counts, memorization, and activity signals, which can rival learned representations on some benchmarks. A central challenge is measuring generalization beyond repeated links and familiar entities, especially under sparse data, long time horizons, and shifts in network structure.

Latest papers

62 papers8 papers in the last four weeks, level with the four weeks before. 0.0% of all new papers.

Never Skip a Batch: Dense Learning of Temporal GNNs via Adaptive Pseudo-Supervision
May 18, 2025Alexander Panyshev, Dmitry Vinichenko, Oleg Travkin +2Temporal GNNs Temporal Link Prediction
Risk AI Research Lab Moscow Independent Research Institute of Artificial Intelligence Moscow, Russia · Risk AI Research Lab Moscow, Russia · Risk AI Research Lab Applied AI Institute Moscow, Russia
FLASH: Flexible Learning of Adaptive Sampling from History in Temporal Graph Neural Networks
Apr 9, 2025Or Feldman, Krishna Sri Ipsit Mantri, Carola-Bibiane Schönlieb +2Temporal GNNs Temporal Link Prediction
Ben-Gurion University of the Negev, Israel · Purdue University, USA · University of Cambridge, UK

May 18, 2025J/K move · Enter open · S save

Never Skip a Batch: Dense Learning of Temporal GNNs via Adaptive Pseudo-Supervision

Alexander Panyshev, Dmitry Vinichenko, Oleg Travkin, Roman Alferov+1

Risk AI Research Lab Moscow Independent Research Institute of Artificial Intelligence Moscow, Russia · Risk AI Research Lab Moscow, Russia · Risk AI Research Lab Applied AI Institute Moscow, Russia

Temporal GNNs Temporal Link Prediction Pseudo-Labeling Semi-Supervised Learning

Open paper AskPDF arXiv

Temporal graph networks suffer from irregular supervision in realworld dynamic graphs, as most minibatches contain few labeled events. The lack of labels leads to high-variance gradient updates and, consequently, slow wall-clock convergence. To constructively reduce sparsity, our Moving-Averaged Labels (MAL) assigns soft pseudo-targets based on past supervised signals using a running label distribution while leaving the loss and the model architecture unchanged. Thus, supervision gaps are replaced with informative signals independent of a temporal graph model and the message passing or memory components used. Theoretical analysis supports our insight that aggregating historical supervision into moving average targets reduces stochastic gradient variance, yielding faster convergence under mild assumptions. Experimentally, for TGNv2 and DyRepv2 (our modification of DyRep) models, MAL boosts predictive performance, establishing a new SOTA, and improves time-to-accuracy (on average 6x faster to reach the top score) for a common suite of Temporal Graph Benchmark datasets.

Latest papers

Never Skip a Batch: Dense Learning of Temporal GNNs via Adaptive Pseudo-Supervision

FLASH: Flexible Learning of Adaptive Sampling from History in Temporal Graph Neural Networks