Building a Deep Neural Network to Automate Protection Against Online Cyber Attacks

Overview

Imagine you are surfing the web, going about your daily business, and suddenly your computer freezes up! Tick... tock... the seconds go by as you wait in anticipation. It is only then that you lift your head towards your camera, only to notice a shining light beside it. You were hacked, somebody is accessing the webcam, and there’s nothing you can do about it except slam your laptop shut. Although not as simple, these sorts of events take place very frequently. In 2021 alone, over 125,000,000 people were victims of fraud attacks! This can happen to anybody at any time — no matter how vigilant you think you are. All it takes is a click of the link, and while there are protection services, an artificial intelligence program that efficiently classifies links as malicious or safe must be created. This project focuses on developing a deep neural network AI that can classify websites as malicious or safe. It does this by parsing any given website for seventy-four identified attributes (such as IP address) and then determining if those seventy-four attributes match a pattern seen among malicious websites. This project was successful and it yielded a 90% accurate AI, that took an average of 26 seconds to output whether a website is safe or not. It is also able to report malicious websites to Google and warn the user through an error box. Future ideas for an expansion project would be looking into phishing emails and increasing AI speed.

Building a Deep Neural Network to Automate Protection Against Online Cyber Attacks

Overview

Competition history

Related projects

Cyber Smart Security: Artificial Intelligence for Network Intrusion Detection Systems

Using Machine Learning To Statically Analyze JavaScript for Cross-Site Scripting Vulnerabilities

Hybrid CNN-LSTM Network Intrusion Detection System Utilizing Multi-Task Learning and Simulated Network Defense

Protection of Deep Neural Networks against Adversarial Attacks with Application to Facial Recognition

Developing a Twitter 'Bot' Identification Application for Public Use

A Novel Machine Learning Approach to Preventing Account Takeovers for Enhanced Cybersecurity

Phishing Website Detection Using Support Vector Machines and Nature-Inspired Optimization Algorithms

Deepfake Detection Using Deep Learning

Browse more like this