Towards privacy: Encrypted deep learning with Syft and Keras
Deep learning need not be irreconcilable with privacy protection. Federated learning enables on-device, distributed model training; encryption keeps model and gradient updates private; differential privacy prevents the training data from...continue reading.