A Cryptographic Cloud Forensics Method For Machine Learning To Increase Security

Abstract

An innovative cryptographic cloud forensics technique designed to improve security in machine learning (ML) settings is presented in the abstract. Data security and confidentiality are becoming increasingly important as cloud-based machine learning systems become more widely used. In an effort to reduce security risks and increase confidence in cloud-based machine learning systems, this technology incorporates cryptography techniques into the forensic investigation process. Delicate information can be handled and broke down securely without compromising protection by using cryptographic procedures, for example, homomorphic encryption and secure multi-party calculation. The suggested method allows for effective forensic investigations in the event of security incidents in addition to providing protection against unauthorized access and data breaches. Both decision trees (DT) and random forests (RF) have accuracy results of 100% for each type of assault detection. The techniques employed for the second phase of data classification were stochastic gradient descent (SGD) learning and logistic regression (LR), both of which produced results of 98% accuracy. What's more, three encryption calculations — rivest figure (RC4), triple information encryption (3DES), and high level encryption standard (AES) — have been utilized to scramble ordered material in light of need. This data will then be securely stored in the cloud.