Computer-implemented method for training a machine learning algorithm

Abstract

The invention relates to a computer-implemented method for training a machine learning algorithm, wherein the machine learning algorithm is used to operate a plurality of devices (10a, 10b, 10c), the method comprising the following steps: - providing and initializing a model to be trained, wherein the model comprises an encoder (16) and a classifier (18); - providing local training data (14a, 14b, 14c) by the devices (10a, 10b, 10c) for local training and providing global training data (14d) by a central processing unit (12) for global training; - performing local training of the encoders (16) by the devices (10a, 10b, 10c); - aggregating the trained encoders (16) by the central processing unit (12); - globally training the classifier (18) by the central processing unit (12);and- Provision of the trained machine learning algorithm by the central computing unit (12).;

Images