arXiv cs.MA· Junyao Zhang, Jinglai Li, Junqi Tang·· 12 小时前AI 评分18
DF-NPE：面向基于智能体模型的双流神经后验估计，减少 90% 直接模拟
Simulation-Efficient Double-Flow Neural Posterior Estimation for Agent-Based Models
AI 导读
研究者提出 Double-Flow Neural Posterior Estimation（DF-NPE），将学习到的时间序列摘要与两个条件归一化流结合，用于合成数据生成和后验估计，并支持池化与有序两种训练策略。
正文
Abstract:Neural posterior estimation (NPE) enables Bayesian inference of the posterior parameter in agent-based models (ABMs) with intractable likelihoods, but simulating training data can be computationally expensive. We propose Double-Flow Neural Posterior Estimation (DF-NPE), which combines learned time-series summaries with two conditional normalizing flows for synthetic-data generation and posterior estimation. The framework supplements direct simulations with synthetic training pairs and supports both pooled and ordered training strategies. Experiments in two settings of the Brock-Hommes asset-pricing model show that DF-NPE can reduce direct simulations by 90 percent while achieving lower mean posterior discrepancies and runtime than standard NPE. Performance varies with the allocation of directly simulated and synthetic data and the training strategy. These results establish learned summary augmentation as a practical approach to reducing simulation costs for Bayesian posterior inference of parameters in ABMs.
| Subjects: | Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2610.35442 [cs.MA] |
| (or arXiv:2610.35442v1 [cs.MA] for this version) | |
| https://doi.org/10.48550/arXiv.2610.35442 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Junqi (Billy) Tang [view email]
[v1]
Mon, 26 Oct 2026 13:41:09 UTC (281 KB)
来源：arXiv cs.MA · arxiv.org