Inspiration

Millions of people are victims of scams every day. Internet crime schemes steal millions of dollars each year from victims and continue to plague the Internet through various methods. Malicious websites are well-known cybersecurity hazards. They are responsible for more than 60% of all cyber-attacks and serve as an efficient method for installing malware, viruses, and other forms of malicious programs online. Malicious URLs can be distributed by email links, SMS messages, browser pop-ups, page advertising, and other means.

SecureNet is a fast and easy way to find out if a website is safe to visit or not. SecureNet will prevent not only adults but also children from accessing harmful websites.

What it does

SecureNet is a browser plugin that leverages machine learning to determine if a website is safe for children or not. If the SercureNet ML model identifies a website as harmful, the user is prevented from accessing it. This will keep children from visiting dangerous websites and will help to avoid cyber attacks. Whenever a user types in a URL, the browser extension sends this URL data to the trained model. The model then predicts if the URL is harmful or not and returns the predicted value. Based on this predicted value the browser extension decided whether the user should be allowed to access the website or not.

How we built it

We collected data that had around a million URLs that were pre-marked as malicious or good. We used this data to train a model to predict if a URL is malicious or not. A logistic regression algorithm was used to train the model with an accuracy of 84%. The entire process of data cleaning to mode deployment was completely automated using the MLRun pipeline. The pipeline for this process can be monitor on the Iguazio platform.

Challenges we ran into

Using the mlrun libraries led us to various errors which we somehow overcome. Also getting used to the Iguazio MLRun platform was a challenge.

Accomplishments that we're proud of

None of us have a background in AI or Machine learning, yet we were able to build the complete pipeline.

What we learned

Understanding ML pipeline. Knowing the difference between DevOps and MLOps. Building prediction model. MLRun, Iguazio platform

What's next for SecureNet

Implementing a full-fledged extension along with user registration. Allowing users to mark websites as dangerous or safe which will help in data collection.

Built With

Share this project:

Updates