Alessandro Palma, Charles Jones, Tristan Cinquin, Fabian J. Theis, Frederik Wenkel · 2026 · arXiv
DOI: 10.48550/arXiv.2610.26749 · open-access PDF
将预先训练的连续时间生成模型适应连续输入的数据,而不忘记以前的任务是终身学习的核心挑战。 一个领先的策略,生成重播,使用一个在历史任务上训练的生成模型,以从经验中增加其自身的连续更新,使用合成样本。 然而,它需要在每个训练步骤中进行完整的轨道模拟,这些模拟会积累分辨错误,并可能降低重播的历史分布。
Machine translation (Meta M2M100); the English original follows.
Adapting pre-trained continuous-time generative models to sequentially incoming data without forgetting earlier tasks is a central challenge in lifelong learning. A leading strategy, generative replay, uses a generative model trained on historical tasks to augment its own continual update with synthetic samples from experience. However, it requires full trajectory simulations at every training step, which accumulate discretization errors and can degrade the replayed historical distribution. Here, we propose trajectory replay, which bypasses sample generation by training an updated velocity fie
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.26749, 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.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange