{"id":"arxiv/2607.03232","url":"https://arxiv.org/abs/2607.03232","tags":["math.CO","stat.ML"],"score":0,"title":"Fast SDP certification of neural networks : towards large multi-class datasets","domain":"math.CO","source":"arxiv","authors":["Margot Boyer","Clément Rambour","Zacharie Alès","Amélie Lambert"],"reviews":0,"subject":"math.CO","summary":"We present a new quadratic model for the certification problem in adversarial robustness, which simultaneously accounts for all possible target classes. Building on this model, we propose a novel semidefinite programming (SDP) relaxation for incomplete verification. A key advantage of our approach is that it certifies robustness in a single optimization, avoiding the need for a separate resolution per class. This yields a significant computational speed-up and enables scalability to large datasets with many classes. To further improve efficiency, we also propose an effective pruning strategy…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-07-03","reactions":{}}