Model Privacy Preservation in Distributed Systems through Machine Learning

The increasing integration of machine learning models into distributed systems raises a pressing concern: how to safeguard privacy of the data and secure the model during sharing. In response to the challenge, a robust solution was proposed. The project's central problem is ensuring the privacy and security of machine learning models within distributed environments. To solve the issue, a machine learning based model is developed for predictive tasks while emphasizing the paramount importance of data privacy and security. With dataset selection, a k-nearest neighbors (KNN) model was constructed. For secure sharing of the model across distributed systems, the Advanced Encryption Standard (AES) algorithm was used to encrypt the model feature and dataset. AES, being a symmetric key algorithm, necessitates secure encryption key exchange. The Rivest-Shamir-Adleman (RSA) algorithm is used to solve the problem, an asymmetric key system with public and private keys. The AES encryption key is securely transmitted via the recipient's public key encryption. Upon reception, decryption enables the reconstruction and utilization of the machine learning based model for further predictions. The project offers a comprehensive solution to the overarching problem of preserving data privacy and model security in distributed systems. By combining various encryption algorithms, the confidentiality and integrity of sensitive machine-learning models within distributed environments were ensured.