Year: 2022
Type: article
Abstract: Cyber threats constitute ill-purposed practices aimed at hacking into computer systems and stealing classified data with individuals and with more frequency, a broad spectrum of organizations as possible targets. This study offers a solution by exploring various machine learning algorithms for predictively analyzing and evaluating cyber threats. This research presents a cutting-edge approach to threat analysis in data center networking using the UNSW-NB15 data, with the goal of improving security. Data cleaning, normalization, feature selection, and balancing were carried out to ensure the model achieved its maximum potential. There is a single binary output feature and forty-three input features in the data. The suggested Deep Neural Network (DNN), along with several DL and ML models like BKP, C-Support Vector Machine (C-SVM), and K-Nearest Neighbors (KNN), were tested. The DNN one is head and shoulders above the competition when it comes to threat recognition and categorization, boasting an amazing ACC of 97.93, PRE of 97, REC of 97, and F1-score (F1) of 97.
Institution: Microsoft (United States)
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7142390112