Protocol for the Interaction of Agents Solving the Problem of Distributed Data Analysis

When analyzing data located on distributed sources, multi-agent technologies are used. The fundamental component of the interaction of agents in is their coordination. There are different strategies for performing distributed data analysis. For example, federated learning can be done sequentially, centrally, or decentralized. The coordination of the agents and the communication protocol used for the interaction must be carried out taking into account the learning strategy. This article describes: communication protocols, interactions between agents, and the types of agents used in each of the federated learning strategies.