MODEL UPDATING METHOD, WORKING NODE, AND MODEL UPDATING SYSTEM

摘要

A model updating method, which can provide a flexible, high-efficiency optimization technology for a decentralized distributed AI system. The method may comprise using the interaction of a working node within a neighborhood to update a local model. The method may comprise receiving one or more pieces of information among at least one parameter of a model, a gradient, and the impulse of the model that are transmitted by another working node in the neighborhood, and combining the local model with data so as to update the model.