{"id":"arxiv/2607.03109","url":"https://arxiv.org/abs/2607.03109","tags":["hep-th","cs.LG"],"score":0,"title":"Graph Neural Networks for the Graphical Bootstrap","domain":"hep-th","source":"arxiv","authors":["Rigers Aliaj","Gabriele Dian","Reza Doobary","Paul Heslop"],"reviews":0,"subject":"hep-th","summary":"We study a graph classification problem involving over 20 million graphs, arising from high-order perturbative computations of correlators in planar $\\mathcal{N}=4$ super-Yang--Mills, a model closely related to the theory of the strong nuclear force. We benchmark graph neural networks, including graph transformers, achieving robust generalization to larger graphs with up to $99.996\\%$ ROC AUC. Then, we analyze how the models can be used to gain a computational speedup compared to the traditional graphical bootstrap algorithm, through shrinking the redundant data by up to $85.5\\%$ at the…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-07-03","reactions":{}}