Exploring Security Vulnerabilities of Deep Learning Models by Adversarial Attacks

Figure 1

An example of generated texts by the visual similar word replacement algorithm.