arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2305.09235cs.LG
收藏

合成数据,真实误差:如何(不)发布和使用合成数据

Synthetic data, real errors: how (not) to publish and use synthetic data

University of Cambridge(剑桥大学);Alan Turing Institute(艾伦·图灵研究所)

Boris van Breugel, Zhaozhi Qian, Mihaela van der Schaar

AI总结:

本文揭示将合成数据直接当作真实数据使用会损害下游模型泛化,并提出深度生成集成框架DGE,通过隐式近似生成模型参数后验分布来改善下游训练、评估与不确定性量化,尤其在少数类和低密度区域提升显著。

摘要:

通过生成模型生成合成数据在机器学习社区内外正引起越来越多的关注,它预示着一个数据集可以按个体需求定制的未来。遗憾的是,合成数据通常并不完美,从而可能在下游任务中产生误差。在这项工作中,我们探索了生成过程如何影响下游机器学习任务。我们表明,朴素的合成数据方法——将合成数据当作真实数据来使用——会导致下游模型和分析对真实数据的泛化效果不佳。作为在合成数据场景下实现更好机器学习的第一步,我们引入了深度生成集成(Deep Generative Ensemble,DGE)——一个受深度集成(Deep Ensembles)启发的框架,旨在隐式近似生成过程模型参数上的后验分布。DGE 改善了下游模型训练、评估和不确定性量化,平均而言大幅优于朴素方法。最大的改进出现在原始数据的少数类和低密度区域,这些区域的生成不确定性最大。

英文摘要:

Generating synthetic data through generative models is gaining interest in the ML community and beyond, promising a future where datasets can be tailored to individual needs. Unfortunately, synthetic data is usually not perfect, resulting in potential errors in downstream tasks. In this work we explore how the generative process affects the downstream ML task. We show that the naive synthetic data approach -- using synthetic data as if it is real -- leads to downstream models and analyses that do not generalize well to real data. As a first step towards better ML in the synthetic data regime, we introduce Deep Generative Ensemble (DGE) -- a framework inspired by Deep Ensembles that aims to implicitly approximate the posterior distribution over the generative process model parameters. DGE improves downstream model training, evaluation, and uncertainty quantification, vastly outperforming the naive approach on average. The largest improvements are achieved for minority classes and low-density regions of the original data, for which the generative uncertainty is largest.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑