PySyft is an open-source framework that enables secured, private computations in deep learning, by combining federated learning and differential privacy in a single programming model integrated into different deep learning frameworks such as PyTorch, Keras or TensorFlow.
In this talk, I will present the main functionalities of PySyft and will showcase two main use cases on which I have personally worked on at the National Research Council (CNR) of Pisa within my Master’s Thesis:
1) PySyft for IoT applications – Distributed training of Deep Neural Networks on Raspberry PIs
2) PySyft for research – Reducing the data consumption of federated learning in a mobile scenario via Neural Network quantization.
Speakers: Daniele Gadler