Toward Defensive Letter Design

TL;DR

This paper explores the adversarial vulnerability of letter images, proposing methods to estimate their defensibility and generate more resilient characters using deep learning and GANs.

Contribution

It introduces a novel approach to enhance letter image robustness against adversarial attacks through a deep regression model and GAN-based generation techniques.

Findings

Letters have quantifiable defensibility against adversarial attacks.

A deep regression model can estimate letter defensibility accurately.

GAN-based methods can generate more defensible letter images.

Abstract

A major approach for defending against adversarial attacks aims at controlling only image classifiers to be more resilient, and it does not care about visual objects, such as pandas and cars, in images. This means that visual objects themselves cannot take any defensive actions, and they are still vulnerable to adversarial attacks. In contrast, letters are artificial symbols, and we can freely control their appearance unless losing their readability. In other words, we can make the letters more defensive to the attacks. This paper poses three research questions related to the adversarial vulnerability of letter images: (1) How defensive are the letters against adversarial attacks? (2) Can we estimate how defensive a given letter image is before attacks? (3) Can we control the letter images to be more defensive against adversarial attacks? For answering the first and second questions, we measure the defensibility of letters by employing Iterative Fast Gradient Sign Method (I-FGSM) and then build a deep regression model for estimating the defensibility of each letter image. We also propose a two-step method based on a generative adversarial network (GAN) for generating character images with higher defensibility, which solves the third research question.