One pixel attack for fooling deep neural networks

Presented by

1. Ziheng Chu

2. Minghao Lu

3. Qi Mai

4. Qici Tan

Methodology

We first formalize the generation of adversarial images as an constrained optimization problem. Let [math]\displaystyle{ \bold{x}\lt math\gt be the vectorized form of an image. Let \lt math\gt f_t(\bold{x}) }[/math]