Fair4Free: Generating High-fidelity Fair Synthetic Samples using Data Free Distillation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sikder, Md Fahim, de Leng, Daniel, Heintz, Fredrik
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913527072030720
author Sikder, Md Fahim
de Leng, Daniel
Heintz, Fredrik
author_facet Sikder, Md Fahim
de Leng, Daniel
Heintz, Fredrik
contents This work presents Fair4Free, a novel generative model to generate synthetic fair data using data-free distillation in the latent space. Fair4Free can work on the situation when the data is private or inaccessible. In our approach, we first train a teacher model to create fair representation and then distil the knowledge to a student model (using a smaller architecture). The process of distilling the student model is data-free, i.e. the student model does not have access to the training dataset while distilling. After the distillation, we use the distilled model to generate fair synthetic samples. Our extensive experiments show that our synthetic samples outperform state-of-the-art models in all three criteria (fairness, utility and synthetic quality) with a performance increase of 5% for fairness, 8% for utility and 12% in synthetic quality for both tabular and image datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2410_01423
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Fair4Free: Generating High-fidelity Fair Synthetic Samples using Data Free Distillation
Sikder, Md Fahim
de Leng, Daniel
Heintz, Fredrik
Machine Learning
Artificial Intelligence
This work presents Fair4Free, a novel generative model to generate synthetic fair data using data-free distillation in the latent space. Fair4Free can work on the situation when the data is private or inaccessible. In our approach, we first train a teacher model to create fair representation and then distil the knowledge to a student model (using a smaller architecture). The process of distilling the student model is data-free, i.e. the student model does not have access to the training dataset while distilling. After the distillation, we use the distilled model to generate fair synthetic samples. Our extensive experiments show that our synthetic samples outperform state-of-the-art models in all three criteria (fairness, utility and synthetic quality) with a performance increase of 5% for fairness, 8% for utility and 12% in synthetic quality for both tabular and image datasets.
title Fair4Free: Generating High-fidelity Fair Synthetic Samples using Data Free Distillation
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2410.01423