Newest papers

Efficient Edge-aware Attention Network for Graph Generation

Vincent Jung, Alba Carballo-Castro, Yiming Qin, Lonneke van der Plas, Pascal Frossard · 2026 · arXiv

DOI: 10.48550/arXiv.2610.28618 · open-access PDF

Spanish

Los modelos generadores de gráficos suelen utilizar arquitecturas de transformadores específicas de gráficos para actualizar conjuntamente las características de nodo y borde. Sin embargo, estas arquitecturas no se benefician directamente de las recientes implementaciones de atención consciente del hardware, limitando su escalabilidad. La adaptación es, sin embargo, no trivial: las características de borde y nivel de grafo deben ser inyectadas en las puntuaciones de atención en pareja, las máscaras de gráficos deben ser respetadas, y las actualizaciones de borde dinámicas siguen siendo una fuente importante de uso de la memoria cuadrada.

Machine translation (Meta M2M100); the English original follows.

Abstract

Graph generative models commonly use graph-specific Transformer architectures to jointly update node and edge features. However, these architectures do not directly benefit from recent hardware-aware attention implementations, limiting their scalability. The adaptation is, however, non-trivial: edge and graph-level features must be injected into pairwise attention scores, graph masks must be respected, and dynamic edge updates remain a major source of quadratic memory usage. We propose an edge-aware attention layer for graph generation that incorporates structural information through head-wise

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28618, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange