{"ID":2847905,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.26185","arxiv_id":"2510.26185","title":"Accumulative SGD Influence Estimation for Data Attribution","abstract":"Modern data-centric AI needs precise per-sample influence. Standard SGD-IE approximates leave-one-out effects by summing per-epoch surrogates and ignores cross-epoch compounding, which misranks critical examples. We propose ACC-SGD-IE, a trajectory-aware estimator that propagates the leave-one-out perturbation across training and updates an accumulative influence state at each step. In smooth strongly convex settings it achieves geometric error contraction and, in smooth non-convex regimes, it tightens error bounds; larger mini-batches further reduce constants. Empirically, on Adult, 20 Newsgroups, and MNIST under clean and corrupted data and both convex and non-convex training, ACC-SGD-IE yields more accurate influence estimates, especially over long epochs. For downstream data cleansing it more reliably flags noisy samples, producing models trained on ACC-SGD-IE cleaned data that outperform those cleaned with SGD-IE.","short_abstract":"Modern data-centric AI needs precise per-sample influence. Standard SGD-IE approximates leave-one-out effects by summing per-epoch surrogates and ignores cross-epoch compounding, which misranks critical examples. We propose ACC-SGD-IE, a trajectory-aware estimator that propagates the leave-one-out perturbation across t...","url_abs":"https://arxiv.org/abs/2510.26185","url_pdf":"https://arxiv.org/pdf/2510.26185v1","authors":"[\"Yunxiao Shi\",\"Shuo Yang\",\"Yixin Su\",\"Rui Zhang\",\"Min Xu\"]","published":"2025-10-30T06:45:22Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.AI\"]","methods":"[]","has_code":false}