Poster (Scientific congresses and symposiums)
Likelihood-free inference with an improved cross-entropy estimator
Stoye, Markus; Brehmer, Johann; Louppe, Gilles et al.
2018Machine Learning and the Physical Sciences, NeurIPS 2019
Peer reviewed
 

Files


Full Text
1808.00973.pdf
Author preprint (370.72 kB)
Download

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
Statistics - Machine Learning; Computer Science - Machine Learning; High Energy Physics - Phenomenology; Physics - Data Analysis; Statistics and Probability
Abstract :
[en] We extend recent work (Brehmer, et. al., 2018) that use neural networks as surrogate models for likelihood-free inference. As in the previous work, we exploit the fact that the joint likelihood ratio and joint score, conditioned on both observed and latent variables, can often be extracted from an implicit generative model or simulator to augment the training data for these surrogate models. We show how this augmented training data can be used to provide a new cross-entropy estimator, which provides improved sample efficiency compared to previous loss functions exploiting this augmented training data.
Disciplines :
Physics
Computer science
Author, co-author :
Stoye, Markus
Brehmer, Johann
Louppe, Gilles  ;  Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Big Data
Pavez, Juan
Cranmer, Kyle
Language :
English
Title :
Likelihood-free inference with an improved cross-entropy estimator
Publication date :
02 August 2018
Event name :
Machine Learning and the Physical Sciences, NeurIPS 2019
Event place :
Vancouver, Canada
Event date :
December 14, 2019
Audience :
International
Peer reviewed :
Peer reviewed
Commentary :
8 pages, 3 figures
Available on ORBi :
since 12 August 2018

Statistics


Number of views
52 (7 by ULiège)
Number of downloads
33 (0 by ULiège)

Bibliography


Similar publications



Contact ORBi