Machine learning method, machine learning device, and machine learning program

The invention provides a machine learning method, a machine learning apparatus, and a machine learning program. The machine learning method acquires a state variable including a processing condition and a physical quantity related to a particle state of a resin, calculates a reward for a determination result of the processing condition based on the state variable, updates a function for determining the processing condition from the state variable based on the calculated reward, and determines the processing condition by repeatedly performing the function update. Therefore, the processing condition for obtaining most rewards is determined. Therefore, the appropriate processing condition of the required particle state of the resin can be easily determined without relying on the experience of skilled technicians all the year round. 本发明提供机器学习方法、机器学习装置以及机器学习程序。机器学习方法获取包含有关树脂的颗粒状态的物理量和处理条件的状态变量，基于状态变量，计算出对处理条件的决定结果的奖励，基于计算出的奖励更新用于根据状态变量决定处理条件的函数，通过反复进行函数更新，从而决定获得奖励最多的处理条件。据此，不依靠熟练的技术人员的长年的经验，也能容易地决定对所要求的树脂的颗粒状态的适当的