TANGO: Graph Neural Dynamics via Learned Energy and Tangential Flows

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Eliasof, Moshe, Haber, Eldad, Schönlieb, Carola-Bibiane
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911095406460928
author Eliasof, Moshe
Haber, Eldad
Schönlieb, Carola-Bibiane
author_facet Eliasof, Moshe
Haber, Eldad
Schönlieb, Carola-Bibiane
contents We introduce TANGO -- a dynamical systems inspired framework for graph representation learning that governs node feature evolution through a learned energy landscape and its associated descent dynamics. At the core of our approach is a learnable Lyapunov function over node embeddings, whose gradient defines an energy-reducing direction that guarantees convergence and stability. To enhance flexibility while preserving the benefits of energy-based dynamics, we incorporate a novel tangential component, learned via message passing, that evolves features while maintaining the energy value. This decomposition into orthogonal flows of energy gradient descent and tangential evolution yields a flexible form of graph dynamics, and enables effective signal propagation even in flat or ill-conditioned energy regions, that often appear in graph learning. Our method mitigates oversquashing and is compatible with different graph neural network backbones. Empirically, TANGO achieves strong performance across a diverse set of node and graph classification and regression benchmarks, demonstrating the effectiveness of jointly learned energy functions and tangential flows for graph neural networks.
format Preprint
id arxiv_https___arxiv_org_abs_2508_05070
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TANGO: Graph Neural Dynamics via Learned Energy and Tangential Flows
Eliasof, Moshe
Haber, Eldad
Schönlieb, Carola-Bibiane
Machine Learning
We introduce TANGO -- a dynamical systems inspired framework for graph representation learning that governs node feature evolution through a learned energy landscape and its associated descent dynamics. At the core of our approach is a learnable Lyapunov function over node embeddings, whose gradient defines an energy-reducing direction that guarantees convergence and stability. To enhance flexibility while preserving the benefits of energy-based dynamics, we incorporate a novel tangential component, learned via message passing, that evolves features while maintaining the energy value. This decomposition into orthogonal flows of energy gradient descent and tangential evolution yields a flexible form of graph dynamics, and enables effective signal propagation even in flat or ill-conditioned energy regions, that often appear in graph learning. Our method mitigates oversquashing and is compatible with different graph neural network backbones. Empirically, TANGO achieves strong performance across a diverse set of node and graph classification and regression benchmarks, demonstrating the effectiveness of jointly learned energy functions and tangential flows for graph neural networks.
title TANGO: Graph Neural Dynamics via Learned Energy and Tangential Flows
topic Machine Learning
url https://arxiv.org/abs/2508.05070