Guardado en:
Detalles Bibliográficos
Autores principales: Mayer, Paul, Luzi, Lorenzo, Siahkoohi, Ali, Johnson, Don H., Baraniuk, Richard G.
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2405.13977
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910631790116864
author Mayer, Paul
Luzi, Lorenzo
Siahkoohi, Ali
Johnson, Don H.
Baraniuk, Richard G.
author_facet Mayer, Paul
Luzi, Lorenzo
Siahkoohi, Ali
Johnson, Don H.
Baraniuk, Richard G.
contents Generative models unfairly penalize data belonging to minority classes, suffer from model autophagy disorder (MADness), and learn biased estimates of the underlying distribution parameters. Our theoretical and empirical results show that training generative models with intentionally designed hypernetworks leads to models that 1) are more fair when generating datapoints belonging to minority classes 2) are more stable in a self-consumed (i.e., MAD) setting, and 3) learn parameters that are less statistically biased. To further mitigate unfairness, MADness, and bias, we introduce a regularization term that penalizes discrepancies between a generative model's estimated weights when trained on real data versus its own synthetic data. To facilitate training existing deep generative models within our framework, we offer a scalable implementation of hypernetworks that automatically generates a hypernetwork architecture for any given generative model.
format Preprint
id arxiv_https___arxiv_org_abs_2405_13977
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Improving Fairness and Mitigating MADness in Generative Models
Mayer, Paul
Luzi, Lorenzo
Siahkoohi, Ali
Johnson, Don H.
Baraniuk, Richard G.
Machine Learning
68T07
Generative models unfairly penalize data belonging to minority classes, suffer from model autophagy disorder (MADness), and learn biased estimates of the underlying distribution parameters. Our theoretical and empirical results show that training generative models with intentionally designed hypernetworks leads to models that 1) are more fair when generating datapoints belonging to minority classes 2) are more stable in a self-consumed (i.e., MAD) setting, and 3) learn parameters that are less statistically biased. To further mitigate unfairness, MADness, and bias, we introduce a regularization term that penalizes discrepancies between a generative model's estimated weights when trained on real data versus its own synthetic data. To facilitate training existing deep generative models within our framework, we offer a scalable implementation of hypernetworks that automatically generates a hypernetwork architecture for any given generative model.
title Improving Fairness and Mitigating MADness in Generative Models
topic Machine Learning
68T07
url https://arxiv.org/abs/2405.13977