发表机构
University of Bern; Inselspital, Bern University Hospital; Graduate School for Cellular and Biomedical Sciences (GCB); Diabetes Center Berne（伯尔尼大学; 伯尔尼大学附属 Inselspital 医院; 细胞与生物医学科学研究生院; 伯尔尼糖尿病中心）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对源分布估计问题，提出基于期望最大化的后验平均方法，通过两种参数化方式实现，在三个基准任务上优于现有固定代理方法。
AI 中文摘要
基于模拟的科学研究通常需要估计模拟器参数的分布，使得该分布的前向映射能够复现一组真实观测值，这就是源分布估计（Source Distribution Estimation, SDE）问题。现有方法通过从固定先验提议训练一次的似然代理来拟合源分布，其目标仅基于该代理而非真实模拟器，因此在参数空间中代理未训练的不准确区域可能失效。我们转而采用期望最大化算法求解SDE问题：E步从当前源估计的新模拟数据训练摊销后验，M步将源重新拟合为该后验在观测数据上的平均值。我们给出两种参数化方式：（1）独立的源流与后验流；（2）单一共享条件流。我们在三个基准任务、宽泛初始先验和错误设定初始先验两种场景下评估该方法，其性能优于现有固定代理方法及各方法的迭代变体，在Lotka-Volterra任务上提升最为显著：所有基线方法在四种初始先验设置中均未达到0.96的数据空间C2ST，而我们的方法在其中三种设置中达到0.64-0.68。
英文摘要
Simulation-based science often requires a distribution over simulator parameters whose push-forward reproduces a set of real observations: this is the source distribution estimation (SDE) problem. Existing methods fit the source against a likelihood surrogate trained once from a fixed proposal prior. Their objective is therefore stated only in terms of the surrogate instead of the true simulator, which may fail for inaccurate areas in parameter space where the surrogate was never trained. We instead solve SDE by expectation maximization: an E-step trains an amortized posterior on fresh simulations from the current source estimate, and an M-step refits the source to the average of that posterior over the observed data. We give two parameterizations, (1) separate source and posterior flows and (2) a single shared conditional flow. We evaluate our method on three benchmark tasks under both broad and misspecified initial priors. Both improve on existing fixed surrogate approaches and on iterated variants of each, most clearly on Lotka--Volterra, where no baseline falls below 0.96 data-space C2ST while our methods reach 0.64-0.68 in three of four initial-prior settings.
