Authors:
Srinivasa Gopi Kumar Peddireddy

Addresses:
Department of Network Implementations and Operations, Charter Communications, Hutto, Texas, United States of America.                                         

Abstract:

Cybersecurity is a fast-changing field nowadays, with organizations facing various attacks that update daily. Traditional methods cannot identify complex threats, so experts use deep learning algorithms to enhance threat detection. This paper explains how deep learning models can be made part of cyber products and the capacity to identify intricate patterns, anomalies, and malicious activities. We demonstrate superior detection efficacy by employing convolutional neural networks (CNNs), recurrent neural networks (RNNs), and hybrid models. The model’s architecture is designed to efficiently handle heavy network traffic, user activity, and application logs. The paper uses the CIC-IDS2017 dataset, an industry-standard cybersecurity benchmark, for training and evaluating models. Libraries like TensorFlow and PyTorch were utilized for training models, whereas the system was coded in high-performance computing environments for scalability testing. Compared to conventional approaches, our findings indicate the potential of deep learning for detecting zero-day attacks and polymorphic malware. Real-world experiences are described in the paper on model selection, hyperparameter optimization, and deployment trends, thus easily deployable in real systems. The study highlights the revolutionary impact deep learning has had on threat detection and cybersecurity infrastructure.

Keywords: Deep Learning; Threat Detection; Neural Networks; Anomaly Detection Data Pre-processing; Convolutional Neural Network (CNN); Recurrent Neural Network (RNN); Distributed Denial-of-Service (DDoS).

Received on: 05/05/2024, Revised on: 11/07/2024, Accepted on: 19/09/2024, Published on: 14/12/2024

DOI: 10.69888/FTSIN.2024.000288

FMDB Transactions on Sustainable Intelligent Networks, 2024 Vol. 1 No. 4, Pages: 190-200

• Views : 37
• Downloads : 5

Download PDF