SecureNet DL: Detecting and Defending Against Al Adversarial Attacks

Author(s)	karthick V, Janakiraman S
Country	India
Abstract	Abstract We introduce a robust deep learning framework that detects and mitigates adversarial attacks using a two-module approach. The first module uses feature-based analysis and statistical methods to detect adversarial inputs, while the second combines adversarial training, feature squeezing, and generative adversarial networks (GANs) to defend against attacks. Evaluated on a CNN with FGSM and PGD attacks on MNIST, the model’s accuracy dropped from 98% to 20–30% under attack, but recovery measures restored performance to 85–90%. This framework enhances AI resilience, supporting safer deployment in sectors like finance, healthcare, and autonomous systems.
Keywords	Adversarial Robustness, Deep Neural Networks, CNN, FGSM, PGD, Defensive Learning, Secure AI.
Field	Computer Applications
Published In	Volume 16, Issue 2, April-June 2025
Published On	2025-06-10
DOI	https://doi.org/10.71097/IJSAT.v16.i2.6097
Short DOI	https://doi.org/g9pz8n

About IJSAT Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us	Message on WhatsApp	+91-9687-182-185	editor@ijsat.org

International Journal on Science and Technology