VARIANTS FOR IMPLEMENTING MACHINE LEARNING WITH REINFORCEMENT IN THE ANYLOGIC PROGRAM

Abstract

Artificial intelligence methods are widely applied in the detection of type problems. There are many variations of machine learning, among which reinforcement learning occupies a special place. At present, there is no answer to the teacher's question, but there is a response from the environment. Such environments can be real (for example, on random roads, in confined airspace, or on a training assembly line) or natural. Simulation software tools allow you to create realistic artificial intelligence environments, safely train and test learning agents. Including such approaches to solving various business problems. This article is devoted to the analysis of implementation options for a coupling simulation model with machine learning with reinforcement, used in the proposed software product, to identify their features, shortcomings and shortcomings.