How to build a trained model and a design support device using the trained model

摘要

PROBLEM TO BE SOLVED: To facilitate setting of a reward in a method of constructing a trained model using reinforcement learning and a design support device using the trained model. SOLUTION: A reward R is given to a combination of a state S determined depending on whether or not a design activity is executed and an action A which is an activity selectable under the state, and the value is maximized. This is a method of constructing a trained model in which the trained model 50 is constructed, that is, step ST4 in which a reward is given based on a pre-input rule 16, and step ST5 in which reinforcement learning is performed based on the given reward. Including, the rule contains information on the source activity, the route destination activity to be performed later, and the thickness indicating the importance of the route destination activity to be performed after the route source activity, and in the step of awarding the reward. When the action immediately before reaching the state matches the route source activity and the action and the route destination activity match, the reward is set based on the thickness. [Selection diagram] FIG. 13