Book Image

Hands-On Artificial Intelligence for Cybersecurity

By : Alessandro Parisi
Book Image

Hands-On Artificial Intelligence for Cybersecurity

By: Alessandro Parisi

Overview of this book

Today's organizations spend billions of dollars globally on cybersecurity. Artificial intelligence has emerged as a great solution for building smarter and safer security systems that allow you to predict and detect suspicious network activity, such as phishing or unauthorized intrusions. This cybersecurity book presents and demonstrates popular and successful AI approaches and models that you can adapt to detect potential attacks and protect your corporate systems. You'll learn about the role of machine learning and neural networks, as well as deep learning in cybersecurity, and you'll also learn how you can infuse AI capabilities into building smart defensive mechanisms. As you advance, you'll be able to apply these strategies across a variety of applications, including spam filters, network intrusion detection, botnet detection, and secure authentication. By the end of this book, you'll be ready to develop intelligent systems that can detect unusual and suspicious patterns and attacks, thereby developing strong network security defenses using AI.
Table of Contents (16 chapters)
Free Chapter
1
Section 1: AI Core Concepts and Tools of the Trade
4
Section 2: Detecting Cybersecurity Threats with AI
8
Section 3: Protecting Sensitive Information and Assets
12
Section 4: Evaluating and Testing Your AI Arsenal

Network attack via model substitution

An interesting demonstration of the potential offered by adversarial attacks conducted in black-box mode is the one described in the paper Practical Black-Box Attacks against Machine Learning (arXiv: 1602.02697v4), in which the possibility of carrying out an attack against remotely hosted DNNs is demonstrated, without the attacker being aware of the configuration characteristics of the target NN.

In these cases, the only information available to the attacker is that of the output returned by the neural network based on the type of input provided by the attacker. In practice, the attacker observes the classification labels returned by the DNN in relation to the attacking inputs. And it is here that an attack strategy becomes interesting. A local substitute model is, in fact, trained in place of the remotely hosted NN, using inputs synthetically...