M.V.N.S.L SUKANYA 1 SK.SHABAZ2 03 07 With the rapid growth of digital communication cloud computing and the Internet of Things IoT protecting sensitive information has become increasingly important Conventional cryptographic algorithms such as AES RSA and ECC provide strong security but face challenges related to computational complexity key management and adaptability against evolving cyber threats Recent advances in Artificial Intelligence AI particularly Neural Networks NNs have opened new opportunities for developing intelligent cryptographic systems capable of learning complex data transformations This research proposes a Neural NetworkBased Encryption and Decryption Algorithm that utilizes deep learning techniques to generate secure encryption mappings while maintaining accurate decryption capabilities The proposed model employs a multilayer neural network trained on plaintextciphertext pairs to learn nonlinear encryption functions The system incorporates dynamic key generation neural weight optimization and secure decoding mechanisms to improve confidentiality and resistance against statistical bruteforce and differential attacks The proposed methodology includes data preprocessing neural network training encryption using learned weights and decryption through inverse neural mapping Performance is evaluated using encryption time decryption accuracy entropy avalanche effect key sensitivity and resistance against cryptographic attacks Experimental results demonstrate that the neural networkbased approach provides enhanced security adaptability and scalability while maintaining competitive computational efficiency compared to traditional encryption techniques Copyright (c) Journal of Science Engineering Technology and Management Science. All rights reserved 2026