Course Description

Machine Learning has proven immensely effective in a diverse set of applications. This trend has reached a new high with the application of Deep Learning virtually in any application domain. This course studies the applications of Machine Learning in the sub domain of Cybersecurity by introducing a plethora of case studies including anomaly detection in networks and computing, side-channel analysis, user authentication and biometrics etc. These case studies are discussed in detail in class, and further examples of potential applications of Machine Learning techniques including Deep Learning are outlined. The course has a strong hands-on component, i.e. students are given datasets of specific security applications and are required to perform simulations.

Prerequisites: Basic understanding of principles of cybersecurity, familiarity with Machine Learning techniques and tools (Matlab or python, numpy, scikit-learn) will be useful but not required.

Textbook

There is no set textbook in this class. Students may consult online tutorials on Machine Learning. During the classes students will be handed out a broad set of reading and discussion materials.

Tentative Course Outline

The following topics will be discussed in class with students in an interactive manner. Each Topic will discuss the material covered in relevant papers which will be handed out at the beginning of the class. Here are some sample topics:

· User Spoofing Identification is Access Control

· Fake Account/News Detection in Social Networks

· Cache Attacks: Classification of Sensitive Information

· Intrusion and Network Anomaly Detection

· Adversarial Learning: Poisoning, Fabrication, etc.

Projects

The students will be assigned with bi-weekly programming/simulation projects in parallel with the course content. Students will be handed datasets and description of a Machine Learning tasks. The goal will be to repeat the cybersecurity objectives as discussed in class on the given dataset.

Final Presentation

The last four weeks are reserved for student presentations. Each student will be required to give an in-class presentation of a paper describing a Machine Learning application in Cybersecurity.

Grading

Grading is based on 5 projects and on an in-class presentation.

The weights for the final grade are as follows:

Projects 75%, Final Presentation 25% (pass/fail).

The following grading scale will be used:

>85% A, 75-84% B, 65-74% C, 55-65% D, else F.