PQMass: Probabilistic Assessment of the Quality of Generative Models using Probability Mass Estimation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lemos, Pablo, Sharief, Sammy, Malkin, Nikolay, Salhi, Salma, Stone, Connor, Perreault-Levasseur, Laurence, Hezaveh, Yashar
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911146064216064
author Lemos, Pablo
Sharief, Sammy
Malkin, Nikolay
Salhi, Salma
Stone, Connor
Perreault-Levasseur, Laurence
Hezaveh, Yashar
author_facet Lemos, Pablo
Sharief, Sammy
Malkin, Nikolay
Salhi, Salma
Stone, Connor
Perreault-Levasseur, Laurence
Hezaveh, Yashar
contents We propose a likelihood-free method for comparing two distributions given samples from each, with the goal of assessing the quality of generative models. The proposed approach, PQMass, provides a statistically rigorous method for assessing the performance of a single generative model or the comparison of multiple competing models. PQMass divides the sample space into non-overlapping regions and applies chi-squared tests to the number of data samples that fall within each region, giving a p-value that measures the probability that the bin counts derived from two sets of samples are drawn from the same multinomial distribution. PQMass does not depend on assumptions regarding the density of the true distribution, nor does it rely on training or fitting any auxiliary models. We evaluate PQMass on data of various modalities and dimensions, demonstrating its effectiveness in assessing the quality, novelty, and diversity of generated samples. We further show that PQMass scales well to moderately high-dimensional data and thus obviates the need for feature extraction in practical applications.
format Preprint
id arxiv_https___arxiv_org_abs_2402_04355
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle PQMass: Probabilistic Assessment of the Quality of Generative Models using Probability Mass Estimation
Lemos, Pablo
Sharief, Sammy
Malkin, Nikolay
Salhi, Salma
Stone, Connor
Perreault-Levasseur, Laurence
Hezaveh, Yashar
Machine Learning
Artificial Intelligence
Methodology
We propose a likelihood-free method for comparing two distributions given samples from each, with the goal of assessing the quality of generative models. The proposed approach, PQMass, provides a statistically rigorous method for assessing the performance of a single generative model or the comparison of multiple competing models. PQMass divides the sample space into non-overlapping regions and applies chi-squared tests to the number of data samples that fall within each region, giving a p-value that measures the probability that the bin counts derived from two sets of samples are drawn from the same multinomial distribution. PQMass does not depend on assumptions regarding the density of the true distribution, nor does it rely on training or fitting any auxiliary models. We evaluate PQMass on data of various modalities and dimensions, demonstrating its effectiveness in assessing the quality, novelty, and diversity of generated samples. We further show that PQMass scales well to moderately high-dimensional data and thus obviates the need for feature extraction in practical applications.
title PQMass: Probabilistic Assessment of the Quality of Generative Models using Probability Mass Estimation
topic Machine Learning
Artificial Intelligence
Methodology
url https://arxiv.org/abs/2402.04355