AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

HNPE: Leveraging Global Parameters for Neural Posterior Estimation

Pedro L. C. Rodrigues, Thomas Moreau, Gilles Louppe, Alexandre Gramfort

NeurIPS 2021 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Simulation-based inference · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add auxiliary observations sharing the global parameterHNPE on the product model · Motivating exampleWasserstein distance of the global-parameter posterior to ground truthImproved
Add auxiliary observations sharing the gainHNPE on the neural mass model · Simulated neural mass modelPosterior concentration around true parametersImproved
Use HNPE instead of the naive stacked-context architectureNeural mass model · Simulated neural mass model, N > 0Normalized Wasserstein distance to ground truthImproved
Use HNPE instead of the aggregation architectureNeural mass model · Simulated neural mass model, N > 0Normalized Wasserstein distance to ground truthImproved
Use the aggregation architecture instead of HNPE (ablation)Neural mass model · Simulated neural mass modelNormalized Wasserstein distance to ground truthNo clear effect
Replace fixed log-PSD summaries with learned YuleNet summaries (ablation)HNPE · Neural mass modelPosterior estimation performanceNo clear effect
Increase the number of SNPE-C refinement rounds (ablation)HNPE · Motivating exampleSinkhorn divergence to the analytic posteriorImproved
Add auxiliary observationsLFVI · Motivating exampleLocal-parameter posterior estimationHurt
Use hierarchical ABC instead of HNPEMotivating example · Motivating examplePosterior concentrationHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy