Artificial Intelligence

    

Learned Data Augmentation with VQ-VAE

Authors: Arnav Dantuluri

In this paper, I propose a simple and easily reproducible method to enhance and extend datasets from as few as 1000 images to as much as 10000 or in essence as many as the user requires. My approach combines a proper latent space modeling of the VAE which is then modified using a process called vector quantization. With these techniques along with enhancing model parameterization and training a simple convolutional neural network can achieve accuracies of up to 93% on synthetic data which proves extremely helpful especially when handling datasets with very few images.

Comments: 9 Pages. (Author's name added to article as required by the rules of viXra.org)

Download: PDF

Submission history

[v1] 2022-03-24 23:11:44

Unique-IP document downloads: 345 times

Vixra.org is a pre-print repository rather than a journal. Articles hosted may not yet have been verified by peer-review and should be treated as preliminary. In particular, anything that appears to include financial or legal advice or proposed medical treatments should be treated with due caution. Vixra.org will not be responsible for any consequences of actions that result from any form of use of any documents on this website.

Add your own feedback and questions here:
You are equally welcome to be positive or negative about any paper but please be polite. If you are being critical you must mention at least one specific error, otherwise your comment will be deleted as unhelpful.

comments powered by Disqus