MINI-BATCH LEARNING DEVICE, OPERATING PROGRAM FOR MINI-BATCH LEARNING DEVICE, OPERATING METHOD FOR MINI-BATCH LEARNING DEVICE, AND IMAGE PROCESSING DEVICE

摘要

Provided are: a mini-batch learning device that can suppress reductions in the accuracy with which a machine learning model that is for performing semantic segmentation distinguishes classes; an operating program and an operating method for the mini-batch learning device; and an image processing device. According to the present invention, when an operating program is run, a CPU of a mini-batch learning device functions as a calculation unit, an identification unit, and an updating unit. The calculation unit calculates an area fraction of mini-batch data for each of a plurality of classes. The identification unit identifies rare classes that have an area fraction below a set value. When the identification unit has identified a rare class, the updating unit reduces the degree to which the machine learning model is updated as compared to when the identification unit has not identified a rare class.