Machine Learning implementation of website fingerprinting
This is a project we worked on for our Machine Learning course. The project attempted to distinguish websites only by viewing timestamp vs. packetsize data. Since this has been successful, but never tested on data where websites overlap.
We took a different approach then the paper presented at: https://www.researchgate.net/publication/306097739_Website_Fingerprinting_at_Internet_Scale. This was an attempt to see if another approach may work better.
It did not provide promising results, but is an ongoing project.
Clone this repository as is, but download the datasets yourself and drag into data folder
Make sure to expand the datasets into the data folder so that you see three folders labeled
- 10s_website1
- 20s_website1
- result
These all will be ignored for a git push in case we want to leave guidelines for any specific things to do within that folder