Impact of Homomorphic Encryption on the Performance of Machine Learning Algorithms

The evolution of the internet and the popularization of access to high-speed connections increased the need for data sharing, especially between business partners. For this reason, the importance of secure data communication, storage, and processing grows. New encryption techniques, such as homomorphic encryption, have been studied in this context, allowing companies to share and analyze data without violating data privacy laws. Nonetheless, it is important to consider potential impacts or overhead associated with these techniques. In this study, we aim to examine the use of machine learning (ML) algorithms on homomorphically-encrypted data. We compare four ML algorithms and analyze the impact of encryption on performance (i.e., in terms of accuracy, precision, recall, and F1-Score) and processing time using a health dataset available on the Kaggle platform. Our analysis demonstrates that it is possible to use ML on data encrypted with homomorphic techniques without significant performance loss. However, it is important to consider the trade-off of longer processing times associated with ML-based solutions working with encrypted data.