Afro-MNIST: Synthetic generation of MNIST-style datasets for low-resource languages

TL;DR

Afro-MNIST introduces synthetic MNIST-style datasets for four African orthographies, enabling machine learning applications in low-resource languages and providing a method for generating such datasets from minimal examples.

Contribution

The paper presents Afro-MNIST datasets for four African scripts and a novel method to generate synthetic MNIST-style datasets from single examples.

Findings

Datasets serve as drop-in replacements for MNIST.

Open-source dataset generation method from single examples.

Supports machine learning education in underrepresented languages.

Abstract

We present Afro-MNIST, a set of synthetic MNIST-style datasets for four orthographies used in Afro-Asiatic and Niger-Congo languages: Ge`ez (Ethiopic), Vai, Osmanya, and N'Ko. These datasets serve as "drop-in" replacements for MNIST. We also describe and open-source a method for synthetic MNIST-style dataset generation from single examples of each digit. These datasets can be found at https://github.com/Daniel-Wu/AfroMNIST. We hope that MNIST-style datasets will be developed for other numeral systems, and that these datasets vitalize machine learning education in underrepresented nations in the research community.