Neural Networks for vulnerability detection -- 2

Using CNN or LSTM, develop a neural network application using python to detect vulnerabilities. Diverse machine learning techniques such as Deep Neural Networks have been successfully applied to malware static analysis. By directly using the source code as input, or the opcodes derived from it, a classifier is able to distinguish between benign or malicious software. A similar approach can be applied to the source code or compiled bytecode to detect potential vulnerabilities of the software (https://samate.nist.gov/SRD/).The goal of this project is to develop a software analysis framework based on neural networks, which will effectively detect one or several potential vulnerabilities in Java programs. First, the source code will be translated to bytecode (http://www.drgarbage.com/bytecode-visualizer/). This will then be passed to a Deep Neural networkfor its classification as robust or vulnerable code.