Differentiable Cluster Discovery in Temporal Graphs

ICLR 2026 Conference Submission19942 Authors

Keywords: Temporal Graph, Clustering, Gumbel Softmax

TL;DR: We propose a differentiable cluster assignment framework for temporal graph.

Abstract: Existing temporal graph clustering methods suffer from poor optimization dynamics due to reliance on heuristically initialized cluster assignment distribution without considering the dynamic nature of the evolving graph. The target cluster assignment distribution often conflicts with evolving temporal representations, leading to oscillatory gradients and unstable convergence. Motivated by the need for differentiable and adaptive clustering in dynamic settings, we propose $\textbf{TGRAIL}$ ($\textbf{T}$emporal $\textbf{Gr}$aph $\textbf{A}$lignment and $\textbf{I}$ndex $\textbf{L}$earning), a novel framework for temporal graph clustering based on Gumbel–Softmax sampling. TGRAIL enables discrete cluster assignments while maintaining gradient flow. To ensure stable training, we formulate the clustering objective as an expectation over Monte Carlo samples and show that this estimator is both unbiased and variance-reduced. Furthermore, we incorporate a temporal consistency loss to preserve the order of interactions across time. Extensive experiments on six real-world temporal graph datasets demonstrate that our approach consistently outperforms state-of-the-art baselines, achieving higher clustering accuracy and robustness. Our results validate the effectiveness of jointly optimizing temporal dynamics and discrete cluster assignments in evolving graphs.

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Submission Number: 19942