A vertical federated learning algorithm for classfication problems with gradient-based optimization.
This algorithm is going to train a shared model on the common IDs (aligned by some private entity resolution technique) between vertically partitioned datasets, as shown in the following figure.
The algorithm itself can be described by the following sequence diagram.
Details of the algorithm can be found in the report.
Details of the implementation (with TensorFlow and Keras) can be found under the directory src.