November 2019
Intermediate to advanced
346 pages
9h 36m
English
We begin by importing torch and its datasets, as well as some associated libraries (Step 1). We then import pysyft and hook it into torch (Step 2). We also create virtual environments for the client and server to simulate a real separation of data. In this step, the crypto_provider serves as a trusted third party to be used for encryption and decryption purposes. In Step 3, we define a simple neural network and, in Step 4, we load-in its pretrained weights. Note that, in Step 5, and, more generally, whenever the .share(...) command is used, you should think of the shared object as being encrypted, and that it is only possible to decrypt it with the assistance of all parties involved. In particular, in Step 9, the test function ...