Learn cryptography encryption and decryption by building convolutional neural networks (CNNs)

• Main Author: Cai, Yu
• Other Authors: Tay Kian Boon
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/165952
• Institution: Nanyang Technological University

In this fast-paced world, digitalization has provided convenience to everyone in all aspects. The trend of research focusing on the implementation of Artificial Intelligence (AI) based cryptography existed thirty years ago. The revolution and the advent of AI and Information and Communication Technology (ICT) have led to a brand-new era of digital reality. New technology is altering the world, with the raising concerns of personal data privacy. Protecting the privacy of everyone’s data is a digital war that should not negligent about. Data security issues are needed to be overcome to prevent personal information from being stolen and misused. The need and demand for more innovative, reliable and usable methods of strengthening cybersecurity are growing, as a result, more research is conducted and in progress. Implementing Convolutional Neural Networks (CNN) in cryptography is widely explored in this decade. Despite the advancements in cryptography, encryption of sensitive data is still a challenging task to fully complete. Most of the recent studies have focused on using neural networks for cryptanalysis and classification accuracy. For this project, the primary objective is to explore the usability of CNN to be implemented in cryptographic operations, which is inclusive of encryption, decryption, and verification. On top of that, achieving a workable, deliverable, and successful algorithm. The results have shown that it is achievable and with huge potential in data protection.