Abstract
- With federated learning scenarios gaining popularity to outsource computational heavy tasks or to increase generalizability of machine learning models, there is also a rise of research in terms of the security and privacy of the respective data used for these tasks. While differential privacy is studied well for Learning Vector Quantization, we want to present steps towards Homomorphic Encryption. In this regard, we will show theoretically how LVQ-1 can be adapted to be compatible with the TFHE encryption scheme and present experimental results.