Distributed machine learning with privacy protection

A system having a plurality of devices that may hosting different versions of an artificial neural network (ANN). In the system, changes to a local version of the ANN may be combined with a main version of the ANN. In the system, a first device may include a memory that may store the main version, a second device may include a memory that may store a local version of the ANN, and there may be a number of devices that store a local version of the ANN. The second device (or any other device of a hosted local version of the system) may include a processor that may train the local version, and a transceiver that may transmit changes to the local version resulting from the training. The first device may include a transceiver that may receive the change in a local version, and a processing device that may combine the received change with the primary version. 一种具有多个装置的系统，所述多个装置可代管人工神经网络(ANN)的不同版本。在所述系统中，所述ANN的本地版本的更改可与所述ANN的主版本组合。在所述系统中，第一装置可包含可存储所述主版本的存储器，第二装置可包含可存储所述ANN的本地版本的存储器，且可存在存储所述ANN的本地版本的许多装置。所述第二装置(或所述系统的