两篇相关论文探讨了生成模型的理论基础，特别关注随机插值。研究分析了这些模型在有限训练数据下的行为，推导了最优场和得分函数的表达式。研究结果表明，生成的样本本质上是添加了噪声的训练样本，其偏差受离散化和估计误差的影响，从而为生成模型中的过拟合和欠拟合提供了新的定义。 AI
影响 为生成模型中的过拟合和欠拟合提供了理论定义，可能指导未来的研究和开发。
排序理由 该集群包含两篇学术论文，讨论了生成模型的理论方面及其训练特性。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2509.21925v2 Announce Type: replace-cross Abstract: This paper investigates the theoretical behavior of generative models under finite training populations. Within the stochastic interpolation generative framework, we derive closed-form expressions for the optimal velocity …
arXiv:2606.08554v1 Announce Type: new Abstract: This paper provides a theoretical account of memorization in stochastic interpolation models. By leveraging closed-form expressions for the optimal velocity field and the associated score function, we show that, in the continuous-ti…
This paper provides a theoretical account of memorization in stochastic interpolation models. By leveraging closed-form expressions for the optimal velocity field and the associated score function, we show that, in the continuous-time oracle setting, both deterministic and stocha…