arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2210.06170stat.MLastro-ph.IMcs.LGhep-ph
收藏

用于基于模拟推断的对比神经比率估计

Contrastive Neural Ratio Estimation for Simulation-based Inference

University of Amsterdam(阿姆斯特丹大学)

Benjamin Kurt Miller, Christoph Weniger, Patrick Forré

AI总结:

本文提出一种无偏的多分类神经比率估计框架,消除了 NRE-B 的固有偏差,统一并推广了 NRE-A 和 NRE-B,并通过多种训练机制下的基准实验给出新的超参数建议和基于互信息上界的评估指标。

摘要:

似然与证据比率估计通常被表述为二分类(NRE-A)或多分类(NRE-B)任务。与二分类框架相比,当前多分类版本的公式存在一个固有的且未知的偏差项,使得原本具有信息量的诊断变得不可靠。我们提出一个在最优情况下不受 NRE-B 固有偏差影响的多分类框架,从而使我们能够运行实践者所依赖的诊断。该框架在一个特殊情形下退化为 NRE-A,在极限情形下退化为 NRE-B。为了公平比较,我们在熟悉和新的训练机制下对所有算法的行为进行了基准测试:当联合抽取的数据不受限制时,当数据固定但先验抽取不受限制时,以及在常见的数据和参数均固定的设置下。我们的研究表明,性能最高的模型在超参数空间中与竞争方法(NRE-A、NRE-B)相距较远。我们给出了不同于先前模型的超参数推荐。我们提出两个互信息上界作为基于模拟推断方法的性能指标,无需后验样本,并提供了实验结果。此版本修正了 $γ$ 中的一个小的实现错误,改进了结果。

英文摘要:

Likelihood-to-evidence ratio estimation is usually cast as either a binary (NRE-A) or a multiclass (NRE-B) classification task. In contrast to the binary classification framework, the current formulation of the multiclass version has an intrinsic and unknown bias term, making otherwise informative diagnostics unreliable. We propose a multiclass framework free from the bias inherent to NRE-B at optimum, leaving us in the position to run diagnostics that practitioners depend on. It also recovers NRE-A in one corner case and NRE-B in the limiting case. For fair comparison, we benchmark the behavior of all algorithms in both familiar and novel training regimes: when jointly drawn data is unlimited, when data is fixed but prior draws are unlimited, and in the commonplace fixed data and parameters setting. Our investigations reveal that the highest performing models are distant from the competitors (NRE-A, NRE-B) in hyperparameter space. We make a recommendation for hyperparameters distinct from the previous models. We suggest two bounds on the mutual information as performance metrics for simulation-based inference methods, without the need for posterior samples, and provide experimental results. This version corrects a minor implementation error in $γ$, improving results.

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。

↑