AI 中文总结
该研究提出模型发现智能体（MDA），结合LLM与贝叶斯机制，在少量干预下发现机制世界模型，在三类基准上实现数据高效模型学习与可靠干预预测的SOTA性能。
AI 中文摘要
预测干预性“如果……会怎样”问题的答案——即从未实施的行动的结果——需要的是**机制性因果模型**，而非曲线拟合；而学习此类模型需要**实验**，因为被动数据无法识别其机制。实验成本高昂，因此核心问题是**数据效率**。我们提出模型发现智能体（Model Discovery Agent, MDA），它将用作候选结构提议者的大语言模型（Large Language Model, LLM）与标准贝叶斯机制相结合：用于参数和结构后验的序贯蒙特卡洛（Sequential Monte Carlo, SMC）、用于难解似然的基于模拟的推理（Simulation-Based Inference, SBI），以及用于实验设计的信息价值（Value-of-Information, VoI），从而从少量干预中发现潜在的机制世界模型。MDA在M-开放设置中运行：当真实情况超出当前假设类时，预测检查会标记出不足，提议者会用新模型扩展假设空间，随后通过设计的实验识别该模型的参数。我们表明，**发现与设计相互强化**：设计步骤识别发现步骤所提议的机制，而识别出的机制可改进预测，从而从剩余未解释的残差中实现进一步发现。在三个不同基准上——涵盖物理学的\textbf{DPbench}（文献[wiemann2026discoverphysics]）、化学的\textbf{CHEMbench}（文献[kabra2026autoscilab]）以及生物学的\textbf{HHbench}（我们创建的新的部分可观测的单神经元电生理学基准）——我们表明，MDA在数据高效模型学习和可靠干预预测能力方面达到了新的SOTA。
英文摘要
A primary goal of science is to learn mechanistic world models from limited experimental data, both to explain observations and to predict novel interventions. We introduce the Model Discovery Agent (MDA), which combines LLM proposals for $\mathcal M$-open model discovery, experiment design based on Value of Information, and approximate Bayesian inference over model structures, parameters, and stochastic latent trajectories. We apply MDA to learn symbolic reaction rate laws for ChemBench \citep{kabra2026autoscilab}, partially observed ODE models for GlucoseBench \citep{xie2018simglucose,kovatchev2009insilico}, and partially observed SDE models for a new stochastic single-neuron simulator we create. In the appendix, we also show results on various other domains from BoxingGym \citep{gandhi2025boxinggym}. We show that MDA has improved sample efficiency compared to various baseline methods, and the learned models are good predictors but also provide interpretable abstractions of each domain.
Commentsv5: Major update: Added GlucoseBench; added experiments with BIC (not just SMC) for marginal likelihood approximation, and de-emphasized SCMC3 narrative; modifed NeuronBench to focus on the stochastic setting; significant rewrite of the main text
