Black-box Adversarial Toolbox (BAT)

A Python Library for Deep Learning Security that focuses on Black-box attacks.

Installation

pip install blackbox-adversarial-toolbox

Usage

import numpy as np
from PIL import Image

from bat.attacks import SimBA
from bat.apis.deepapi import VGG16Cifar10

# Load Image [0.0, 1.0]
x = np.asarray(Image.open('dog.jpg').resize((32, 32))) / 255.0

# Initialize the Cloud API Model
DEEP_API_URL = 'https://api.wuhanstudio.uk'
model = VGG16Cifar10(DEEP_API_URL + "/vgg16_cifar10")

# SimBA Attack
simba = SimBA(model)
x_adv = simba.attack(x, epsilon=0.1, max_it=1000)

# Distributed SimBA Attack
x_adv = simba.attack(x, epsilon=0.1, max_it=1000, distributed=True , batch=50, max_workers=10)