Measuring Performance of Generative Adversarial Networks on Devanagari Script

TL;DR

This paper evaluates the performance of GANs on the complex Devanagari script by developing custom classifiers to assess output quality, moving beyond traditional digit datasets like MNIST.

Contribution

It introduces a novel application of GANs to Devanagari script and proposes a new evaluation method using classifiers for real-world script generation.

Findings

GANs can generate Devanagari script with measurable quality

Custom classifiers effectively evaluate script generation

Devanagari script poses unique challenges for GAN training

Abstract

The working of neural networks following the adversarial philosophy to create a generative model is a fascinating field. Multiple papers have already explored the architectural aspect and proposed systems with potentially good results however, very few papers are available which implement it on a real-world example. Traditionally, people use the famous MNIST dataset as a Hello, World! example for implementing Generative Adversarial Networks (GAN). Instead of going the standard route of using handwritten digits, this paper uses the Devanagari script which has a more complex structure. As there is no conventional way of judging how well the generative models perform, three additional classifiers were built to judge the output of the GAN model. The following paper is an explanation of what this implementation has achieved.