Efficient Estimation of Generative Models Using Tukey Depth
Generative models have recently received a lot of attention. However, a challenge with such models is that it is usually not possible to compute the likelihood function, which makes parameter estimation or training of the models challenging. The most commonly used alternative strategy is called like...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2024-03-01
|
Series: | Algorithms |
Subjects: | |
Online Access: | https://www.mdpi.com/1999-4893/17/3/120 |
_version_ | 1797242346915495936 |
---|---|
author | Minh-Quan Vo Thu Nguyen Michael A. Riegler Hugo L. Hammer |
author_facet | Minh-Quan Vo Thu Nguyen Michael A. Riegler Hugo L. Hammer |
author_sort | Minh-Quan Vo |
collection | DOAJ |
description | Generative models have recently received a lot of attention. However, a challenge with such models is that it is usually not possible to compute the likelihood function, which makes parameter estimation or training of the models challenging. The most commonly used alternative strategy is called likelihood-free estimation, based on finding values of the model parameters such that a set of selected statistics have similar values in the dataset and in samples generated from the model. However, a challenge is how to select statistics that are efficient in estimating unknown parameters. The most commonly used statistics are the mean vector, variances, and correlations between variables, but they may be less relevant in estimating the unknown parameters. We suggest utilizing Tukey depth contours (TDCs) as statistics in likelihood-free estimation. TDCs are highly flexible and can capture almost any property of multivariate data, in addition, they seem to be as of yet unexplored for likelihood-free estimation. We demonstrate that TDC statistics are able to estimate the unknown parameters more efficiently than mean, variance, and correlation in likelihood-free estimation. We further apply the TDC statistics to estimate the properties of requests to a computer system, demonstrating their real-life applicability. The suggested method is able to efficiently find the unknown parameters of the request distribution and quantify the estimation uncertainty. |
first_indexed | 2024-04-24T18:37:46Z |
format | Article |
id | doaj.art-76cd27c839b346e99189e39e528300af |
institution | Directory Open Access Journal |
issn | 1999-4893 |
language | English |
last_indexed | 2024-04-24T18:37:46Z |
publishDate | 2024-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Algorithms |
spelling | doaj.art-76cd27c839b346e99189e39e528300af2024-03-27T13:17:26ZengMDPI AGAlgorithms1999-48932024-03-0117312010.3390/a17030120Efficient Estimation of Generative Models Using Tukey DepthMinh-Quan Vo0Thu Nguyen1Michael A. Riegler2Hugo L. Hammer3Department of Mathematics and Computer Science, VNUHCM—University of Science, District 5, Ho Chi Minh City 70000, VietnamSimula Metropolitan Center for Digital Engineering, 0167 Oslo, NorwaySimula Metropolitan Center for Digital Engineering, 0167 Oslo, NorwaySimula Metropolitan Center for Digital Engineering, 0167 Oslo, NorwayGenerative models have recently received a lot of attention. However, a challenge with such models is that it is usually not possible to compute the likelihood function, which makes parameter estimation or training of the models challenging. The most commonly used alternative strategy is called likelihood-free estimation, based on finding values of the model parameters such that a set of selected statistics have similar values in the dataset and in samples generated from the model. However, a challenge is how to select statistics that are efficient in estimating unknown parameters. The most commonly used statistics are the mean vector, variances, and correlations between variables, but they may be less relevant in estimating the unknown parameters. We suggest utilizing Tukey depth contours (TDCs) as statistics in likelihood-free estimation. TDCs are highly flexible and can capture almost any property of multivariate data, in addition, they seem to be as of yet unexplored for likelihood-free estimation. We demonstrate that TDC statistics are able to estimate the unknown parameters more efficiently than mean, variance, and correlation in likelihood-free estimation. We further apply the TDC statistics to estimate the properties of requests to a computer system, demonstrating their real-life applicability. The suggested method is able to efficiently find the unknown parameters of the request distribution and quantify the estimation uncertainty.https://www.mdpi.com/1999-4893/17/3/120generative modelsTukey depthlikelihood-free estimationcomputer resource management |
spellingShingle | Minh-Quan Vo Thu Nguyen Michael A. Riegler Hugo L. Hammer Efficient Estimation of Generative Models Using Tukey Depth Algorithms generative models Tukey depth likelihood-free estimation computer resource management |
title | Efficient Estimation of Generative Models Using Tukey Depth |
title_full | Efficient Estimation of Generative Models Using Tukey Depth |
title_fullStr | Efficient Estimation of Generative Models Using Tukey Depth |
title_full_unstemmed | Efficient Estimation of Generative Models Using Tukey Depth |
title_short | Efficient Estimation of Generative Models Using Tukey Depth |
title_sort | efficient estimation of generative models using tukey depth |
topic | generative models Tukey depth likelihood-free estimation computer resource management |
url | https://www.mdpi.com/1999-4893/17/3/120 |
work_keys_str_mv | AT minhquanvo efficientestimationofgenerativemodelsusingtukeydepth AT thunguyen efficientestimationofgenerativemodelsusingtukeydepth AT michaelariegler efficientestimationofgenerativemodelsusingtukeydepth AT hugolhammer efficientestimationofgenerativemodelsusingtukeydepth |