During this weeks workshop practical i managed to think up an idea which ticks all the boxes of what i wanted my idea to be, if not quite difficult to pull off.

The idea is to investigate using neural networking (possibly tensorflow from google) and machine learning techniques (MLT) in the context of Red team. This is in the hopes of creating a useful module or AI which can be used by the Red Team in capture the flag events. If successful an extention would be to have a comparison between running it on a GPU cluster (i.e. a large neural cluster vs small neural cluster).

A lot of research will need to be done in order to give me a basic starting point in MLT (sub field of AI) and Neural Networking using TensorFlow (deep neural networking?) in order to actually incorporate them.

Working Title:

An investigation into using neural networking and Machine learning Techniques in the context of Red team.

AIM:

OBJECTIVES:

RESEARCH QUESTION:

once im able to fill those fields ill create a powerpoint for this weeks workshop task.

Since i know practically nothing about MLT and Deep learning ive decided to start with these:

https://www.youtube.com/watch?v=b99UVkWzYTQ

https://www.youtube.com/watch?v=UzxYlbK2c7E&list=PLA89DCFA6ADACE599

googles online tensorflow playground. programming has usually been my strengths although not very experienced with python i dont think itl be the hardest part of the project.

https://en.wikipedia.org/wiki/Deep_learning

http://deeplearning4j.org/ - deep learning library for java

Update 2:

New Working Title:

An Investigation into Deep Neural Networking and Machine Learning Techniques focusing on Red teaming Procedures.

Aim: To investigate artificial deep neural networks and machine learning techniques focusing on Red Teaming procedures.

Objectives:

• To research current deep neural networking trends in the security industry.
• To devise and create, with reasoning, a suitable software implementation of neural networking and machine learning in the context of Red Teaming.
• To analyse and evaluate the effectiveness of Deep neural networks for Red Teaming.
• To compare the training of neural networks on an average desktop workstation versus a GPU cluster. POSSIBILITY
• To make recommendations and suggest future work.

Topic: Neural networking and machine learning in offensive cyber security.

• RQ: Analysing the current state of neural networking in the cyber security sector.
• RQ: Investigating the use and effectiveness of deep neural networks in Red teaming.

There are many different librarys and frameworks for deep neural networking such as deeplearning4j, tensorflow and others. These will need to be researched further and the differences between them made more clear.