NOVAHackathon

Using AI to detect cyber attacks

Name of the challenge: Rigil Cyber Security AI Challenge Situation / Need: The goal of this challenge is to demonstrate an effective use of Machine Learning in identifying malicious behavior in security-related data. Security-related data can include network traffic logs, firewall logs, Windows event logs, etc.
In framing your solution, we encourage you to first read the article below by Cyber Security expert Raffael Marty: https://towardsdatascience.com/ai-and-machine-learning-in-cyber-security-d6fbee480af0 The diagram shown on the Raffael Marty presents a great overview of the types of solutions/algorithms you can use: https://cdn-images-1.medium.com/max/1600/1*a-qa3jgOFsHyABv6U0zaXA.png We encourage you to use one of the data sets under the Datasets heading of the GitHub “Awesome Machine Learning for Cyber Security” below: https://github.com/jivoi/awesome-ml-for-cybersecurity#-datasets You are also welcome to use any of the tutorials or resources hosted on this GitHub page to guide you in creating your solution.

Requirements (what exactly your company wants), Please be specific:

1. Create a novel AI-based algorithm/model/solution which uses one of the Cyber Security data sets to effectively and meaningfully identify malicious activity.

2. Demonstrate the effectiveness of your model/algorithm by applying it to new/test data sets with a high level of accuracy, precision, and recall

3. Explain how your solution has real world applications You are free to use R Studio, Spyder, Jupyter notebooks, a web application, or similar tools to demonstrate your solution. Bonus points if you are able to deploy your solution to a mini web application incorporating data visualization libraries like d3.

4. Be prepared to explain: a. Which of the categories of approaches mentioned in Raffael Marty’s diagram your approach most closely falls under. b. Why you chose the approach you chose, and what is novel/different about it. c. How your solution has real world applications

How would your company score the projects or solutions turned in by the students (point system 1 - 10): In this section, please identify how a group of students would program or structure their solution to fit the judging needs. What is your company looking for, and what is most important for the company?

1. Solution is meaningful/nontrivial and effectively uses Machine Learning to identify the type of threat in a more effective manner than approaches that are not AI-based
2. Solution has high Accuracy, Precision, and Recall when applied to new/test data sets.
3. Uniqueness of solution – how different is the solution from existing solutions or those readily found in online tutorials etc
4. Visual presentation

*Please make sure to turn in this form before April the 10th to tmitchell93@gwmail.gwu.edu.