RansomedLSTM: Ransomware Detection using Recurrent Neural Networks (with M. Tayabb and J. Carlton)

In this paper we have proposed a novel idea for the detection and prevention of ransomwares leveraging the power of deep learning. We trained a Long Short Term Memory (LSTM) Neural Network, to detect the ransomware infection. We gathered the binaries infected with the real world ransomwares and the programs which perform the encryption and decryption legitimately. The trained LSTM is able to classify the ransom-wares accurately.

---

Cross-Site Request Forgery Attack Project

In this paper I succeed in a CSRF attack on a user that is part of a social networking site: Elgg, which is hosted locally on a VM. The goal is to have our user (Boby) become friends with the user (Alice) when Alice clicks on a link to a malicious website. It is assumed that Alice has an active session with Elgg, and that she received the URL to the malicious website via email or a posting on Elgg.

---

A Code Injection Attack on HTML5 Based Mobile Apps

In this paper I explain and demonstrate my code injection attack on an HTML5 based music player app on an Android device. A code injection attack is a type of attack used to manipulate a program or web server with the goal of changing the way it is executed. Code injections require a vulnerability in the program that allows the attacker to input code. This attack is similar to a cross-site scripting attack (a type of code injection), however it differs in one key area. A cross-site scripting attack exploits a vulnerability in a web application, which only has one channel for code injection, through the web server, while my attack on an HTML5 based app could use many different channels for code injection even though I will be focusing mostly on "Contacts". My attack solely takes place on an Android operating system.

---

Related Courses:

• Computer & Network Security (Graduate)
• Security & Privacy in Cloud Computing (Graduate)
• Wireless Network Security & Privacy (Graduate)