Neural Permutation Processes

Ari Pakman; Yueqi Wang; Liam Paninski

Neural Permutation Processes

Ari Pakman, Yueqi Wang, Liam Paninski

Research output: Contribution to journal › Conference article › peer-review

Abstract

We introduce a neural architecture to perform amortized approximate Bayesian inference over latent random permutations of two sets of objects. The method involves approximating permanents of matrices of pairwise probabilities using recent ideas on functions defined over sets. Each sampled permutation comes with a probability estimate, a quantity unavailable in MCMC approaches. We illustrate the method in sets of 2D points and MNIST images.

Original language	English
Journal	Proceedings of Machine Learning Research
Volume	118
State	Published - 1 Jan 2019
Externally published	Yes
Event	2nd Symposium on Advances in Approximate Bayesian Inference, AABI 2019 - Vancouver, Canada Duration: 8 Dec 2019 → …

ASJC Scopus subject areas

Artificial Intelligence
Software
Control and Systems Engineering
Statistics and Probability

Cite this

@article{ecdfdd6b582b493db51ad9241deea714,

title = "Neural Permutation Processes",

abstract = "We introduce a neural architecture to perform amortized approximate Bayesian inference over latent random permutations of two sets of objects. The method involves approximating permanents of matrices of pairwise probabilities using recent ideas on functions defined over sets. Each sampled permutation comes with a probability estimate, a quantity unavailable in MCMC approaches. We illustrate the method in sets of 2D points and MNIST images.",

author = "Ari Pakman and Yueqi Wang and Liam Paninski",

note = "Publisher Copyright: {\textcopyright} A. Pakman, Y. Wang & L. Paninski.; 2nd Symposium on Advances in Approximate Bayesian Inference, AABI 2019 ; Conference date: 08-12-2019",

year = "2019",

month = jan,

day = "1",

language = "English",

volume = "118",

journal = "Proceedings of Machine Learning Research",

issn = "2640-3498",

}

TY - JOUR

T1 - Neural Permutation Processes

AU - Pakman, Ari

AU - Wang, Yueqi

AU - Paninski, Liam

N1 - Publisher Copyright: © A. Pakman, Y. Wang & L. Paninski.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We introduce a neural architecture to perform amortized approximate Bayesian inference over latent random permutations of two sets of objects. The method involves approximating permanents of matrices of pairwise probabilities using recent ideas on functions defined over sets. Each sampled permutation comes with a probability estimate, a quantity unavailable in MCMC approaches. We illustrate the method in sets of 2D points and MNIST images.

AB - We introduce a neural architecture to perform amortized approximate Bayesian inference over latent random permutations of two sets of objects. The method involves approximating permanents of matrices of pairwise probabilities using recent ideas on functions defined over sets. Each sampled permutation comes with a probability estimate, a quantity unavailable in MCMC approaches. We illustrate the method in sets of 2D points and MNIST images.

UR - http://www.scopus.com/inward/record.url?scp=85161346968&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:85161346968

SN - 2640-3498

VL - 118

JO - Proceedings of Machine Learning Research

JF - Proceedings of Machine Learning Research

T2 - 2nd Symposium on Advances in Approximate Bayesian Inference, AABI 2019

Y2 - 8 December 2019

ER -

Neural Permutation Processes

Abstract

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this