Sentinel Sense

Inspiration

Our team’s passion for AI and cybersecurity inspired us to explore innovative ways to enhance digital security. We chose to focus on intrusion detection, aiming to provide a comprehensive analysis that goes beyond just identifying specific activities. By leveraging AI, we can deliver a more comprehensive understanding of potential threats, empowering users with insights that strengthen their overall security posture.

What it does

Users can easily submit their activity log datasets, allowing our model to analyze and identify potential threats. The system will promptly inform them of any detected malicious activities, providing clear insights into the nature of the threats.

How we built it

We developed a program that creates random data to mimic network traffic. After reading data, we ran it through a recurrent neural network with K-Folding to classify portions as either clean or malicious. This model was saved to be later applied to new dataset The frontend utilized Bootstrap and was reconfigured with a Flask Endpoint to accept new CSV files, run the CSV file against the model, and present the results to the user.

Challenges we ran into

Two of our members were new to most of the frameworks used in this system. We ran into overfitting issues with the generated data with the under-tuned model, poor classification of the minority class, and integrating the model onto the endpoint. After increasing the width and depth of the model, we achieved an unbiased solution to new datasets.

Accomplishments that we're proud of

We are proud that we learned new concepts and applications in such a short amount of time. This was the first project finally applying theoretical ideas into practice.

What we learned

This was our front-end engineers' first experience competing in a hackathon, and it marked their initial exposure to HTML, CSS, JavaScript, and Flask. It proved to be an engaging and enlightening opportunity, providing valuable experience in new technologies. Additionally, this project offered our front-end engineers a glimpse into the transformative potential of machine learning and its application in creating solutions that can benefit others. For one of our AI engineers, this was their first time creating a model, applying theoretical knowledge into a proper set. This was the first time for both AI engineers to use Pytorch, which was faster than previous tensor languages.

What's next for Sentinel Sense

Sentinel is excited to further elevate this project. We see substantial opportunities for enhancement, especially in optimizing the front-end for a more effective presentation of results through engaging visualizations. Another goal is to refine the back-end, allowing us to provide users with comprehensive conclusions or summaries after thorough data analysis. To enhance the user experience, we also plan to offer personalized recommendations such as anomaly alerts, security best practices, and incident response plans. Future integration with network traffic will provide security analyst real time insights enabling them to proactively monitor and address potential threats.