Author(s) :

YUVAN KUMAR SALINA.

Volume/Issue :

Abstract :

The exponential expansion of machine learning (ML) applications in a variety of fields has raised concerns about the privacy of user data. Strong privacy-preserving methods must be developed because sensitive data used in model training may unintentionally be revealed. Under the headings of data anonymization, differential privacy, federated learning, homomorphic encryption, and secure multi-party computation, this survey offers a thorough summary of the privacy-preserving techniques currently used in machine learning. We examine their methods, advantages, drawbacks, and potential uses. In order to guarantee privacy compliance in ML-driven systems, the paper also lists the main obstacles, unresolved research issues, and necessary future paths.

No. of Downloads :

0