Homomorphic encryption performance in secure data processing for machine learning

In terms of data privacy and security especially in the development of machine learning HE has the potential to solve it. This paper focuses on homomorphic encryption as a method for secure data processing with reference to encryption techniques that allow arithmetic operations to be performed on encrypted data for use in machine learning. When comparing different HE methods, their computational complexity, as well as the obtained precision, have been analyzed, and practical suitability of the methods in question has been discussed, that is, what actual problem-solving can be done using the methods in question. That was done in a way that does not in any way encroach into the security of the data to ensure that we do not compromise it in the proposed method used to incorporate HE with machine learning models. The provided findings of experiments show that the application of HE in experiments secured data is valid while establishing the cost between security measures and computation time. Chen’s works provide knowledge on how more efficient, personal sensitive preserving machine learning can be developed and integrated.