TechSavvyGranny

A World Powered By AI

Inspiration

Our drive and curiousity for machine learning, cyber security, and data science pushed us to make a complex antivirus with intruistic features beyond our years of experience by simply collaborating together and using the opportunity our parents gave us to make the best of ourselves.

What it does

TechSavvyGranny is machine-learning oriented. Essentially, the software checks files as they are downloaded. If a file is flagged as malicious, it can then be targeted for further examination or eliminated. The most important part of this software is the methods used to obtain an accurate machine learning model. We parsed PE file headers because they conceal a large amount of data about an executable file.

How we built it

Our antivirus was written mainly in python3 and is driven by supervised machine learning. We used Keras, and NumPy to supervise the AI. As an additional feature we used the classic signature method to help increase overall malware detection for our program. We used a dataset with PE (Portable Executable) file headers to train our model to recognize malicious patterns. We also set up multiple monitoring systems for packet analysis, and RAM usage. We used data normalization to increase accuracy and learning speed for the AI.

Challenges we ran into

One of the main challenges we ran into was formatting the .csv dataset the right way for the AI to start progressing properly. Our dataset was extremely big and wasn't formatted in the most efficient way that slowed our learning speed by hours.

Accomplishments that we're proud of

Having a full working AI model with 80% accuracy for malware detection
Successfully learning how to use Keras, and Tensorflow library
In-Depth profiency for detection of different malware features.
Creating such a complex project in 24 hours
Being completely sleep deprived for the whole competition

What we learned

How to successfully implement machine learning libraries
Learned how to use supervised machine learning for malware detection
How to normalize data
Learned to use the Google Cloud Platform for AI development
Python3

What's next for TechSavvyGranny

We hope to bring TechSavvyGranny to the people who need its services most.

Built With

Submitted to

MasseyHacks V

Created by

I made a lot of things. I created a spicy PE header parser. I also made the ML model and its adjacent supporting software, including random parsers and bits of php here and there.

Alexander Chudinov
Software Engineer | UBC Student
I mostly worked on the UI with Java and made it accessible, especially with those with bad vision. Unfortunately, the python integration didn't go well.

CompleteTheSquare
During this project, I worked on our resource monitor for our external device (which we ended up not using) and I also ended up helping Alex with the CVS file parser. Aside from that I also learned a substantial amount of python and machine learning :)

Aryan Afrouzi
48 6f 6e 6b
During this 24 contest I worked on helping Alex with development of the AI, and feeding it the right formatted data. Code design/structure, README file.

Rope Paradox