Generative Expansion of Small Datasets: An Expansive Graph Approach

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jebraeeli, Vahid, Jiang, Bo, Krim, Hamid, Cansever, Derya
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912053813313536
author Jebraeeli, Vahid
Jiang, Bo
Krim, Hamid
Cansever, Derya
author_facet Jebraeeli, Vahid
Jiang, Bo
Krim, Hamid
Cansever, Derya
contents Limited data availability in machine learning significantly impacts performance and generalization. Traditional augmentation methods enhance moderately sufficient datasets. GANs struggle with convergence when generating diverse samples. Diffusion models, while effective, have high computational costs. We introduce an Expansive Synthesis model generating large-scale, information-rich datasets from minimal samples. It uses expander graph mappings and feature interpolation to preserve data distribution and feature relationships. The model leverages neural networks' non-linear latent space, captured by a Koopman operator, to create a linear feature space for dataset expansion. An autoencoder with self-attention layers and optimal transport refines distributional consistency. We validate by comparing classifiers trained on generated data to those trained on original datasets. Results show comparable performance, demonstrating the model's potential to augment training data effectively. This work advances data generation, addressing scarcity in machine learning applications.
format Preprint
id arxiv_https___arxiv_org_abs_2406_17238
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Generative Expansion of Small Datasets: An Expansive Graph Approach
Jebraeeli, Vahid
Jiang, Bo
Krim, Hamid
Cansever, Derya
Machine Learning
Computer Vision and Pattern Recognition
Image and Video Processing
Limited data availability in machine learning significantly impacts performance and generalization. Traditional augmentation methods enhance moderately sufficient datasets. GANs struggle with convergence when generating diverse samples. Diffusion models, while effective, have high computational costs. We introduce an Expansive Synthesis model generating large-scale, information-rich datasets from minimal samples. It uses expander graph mappings and feature interpolation to preserve data distribution and feature relationships. The model leverages neural networks' non-linear latent space, captured by a Koopman operator, to create a linear feature space for dataset expansion. An autoencoder with self-attention layers and optimal transport refines distributional consistency. We validate by comparing classifiers trained on generated data to those trained on original datasets. Results show comparable performance, demonstrating the model's potential to augment training data effectively. This work advances data generation, addressing scarcity in machine learning applications.
title Generative Expansion of Small Datasets: An Expansive Graph Approach
topic Machine Learning
Computer Vision and Pattern Recognition
Image and Video Processing
url https://arxiv.org/abs/2406.17238