Leveraging Global Parameters for Flow-based Neural Posterior Estimation
22 февраля 2021 года
21:37
Leveraging Global Parameters for Flow-based Neural Posterior Estimation
Текст новости:
Title: Leveraging Global Parameters for Flow-based Neural Posterior Estimation
Author, co-author: Rodrigues, Pedro; Moreau, Thomas; Louppe, Gilles; Gramfort, Alexandre
Abstract: Inferring the parameters of a stochastic model based on experimental observations is central to the scientific method. A particularly challenging setting is when the model is strongly indeterminate, i.e., when distinct sets of parameters yield identical observations. This arises in many practical situations, such as when inferring the distance and power of a radio source (is the source close and weak or far and strong?) or when estimating the amplifier gain and underlying brain activity of an electrophysiological experiment. In this work, we present a method for cracking such indeterminacy by exploiting additional information conveyed by an auxiliary set of observations sharing global parameters. Our method extends recent developments in simulation-based inference(SBI) based on normalizing flows to Bayesian hierarchical models. We validate quantitatively our proposal on a motivating example amenable to analytical solutions, and then apply it to invert a well known non-linear model from computational neuroscience.
Связанные объекты: #A (найти в новостях).

Текст со страницы (автоматическое получение):
Leveraging Global Parameters for Flow-based Neural Posterior Estimation
Language :
Abstract :
[en] Inferring the parameters of a stochastic model based on experimental observations is central to the scientific method. A particularly challenging setting is when the model is strongly indeterminate, i.e., when distinct sets of parameters yield identical observations. This arises in many practical situations, such as when inferring the distance and power of a radio source (is the source close and weak or far and strong?) or when estimating the amplifier gain and underlying brain activity of an electrophysiological experiment. In this work, we present a method for cracking such indeterminacy by exploiting additional information conveyed by an auxiliary set of observations sharing global parameters. Our method extends recent developments in simulation-based inference(SBI) based on normalizing flows to Bayesian hierarchical models. We validate quantitatively our proposal on a motivating example amenable to analytical solutions, and then apply it to invert a well known non-linear model from computational neuroscience.
Target :
Автоматическая система мониторинга и отбора информации
Источник
Другие материалы рубрики
★★★★★  22 февраля 2021 года
09:49
HOLISMOKES. III. Achromatic phase of strongly lensed Type Ia supernovae
  18 февраля 2021 года
14:29
Dynamical behavior of alternate base expansions
★★★★  18 февраля 2021 года
11:50
Les difficultés d'apprentissage de la trigonométrie
★★★  17 февраля 2021 года
09:05
Introduction to ecological modelling
★★★  17 февраля 2021 года
09:03
Diagenetic Modelling
★★★★  17 февраля 2021 года
09:01
Dynamical properties of greedy and lazy alternate base expansions
★★★  17 февраля 2021 года
05:00
Научные результаты ОИЯИ за 2020 год
★★★  16 февраля 2021 года
15:55
Electrical signature of Co2-rich groundwater systems
★★★★  16 февраля 2021 года
11:22
A parsimonious urban production - What’up in Wallonia?
★★  15 февраля 2021 года
08:57
Trapped field of two close bulk superconductors with misaligned c-axes