MACHINE LEARNING DEVICE, MACHINE LEARNING METHOD, MACHINE LEARNING PROGRAM, AND INSPECTION DEVICE

• Main Authors: KUROSAWA KIMINORI, NAKATORI SHINICHI
• Format: Patent
• Language: English; Japanese
• Published: 19.03.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIRCHEMICAL OR PHYSICAL PROPERTIES; MEASURING; PHYSICS; TESTING

To realize a machine learning device with which it is possible to perform additional learning of a classifier used for classifying goods before the classification accuracy of the classifier degrades.SOLUTION: A machine learning device 1 for causing machine learning to be exercised by a classifier set (101) consisting of a plurality of classifiers for classifying the states of goods belonging to each of a plurality of groups referring to an image including the goods as a subject, comprises: an evaluation unit (102) for evaluating, for each group, the classification accuracy of each classifier included in the classifier set; and a learning control unit (103) which, when the classification accuracy of at least one classifier included in the classifier set does not reach a predetermined standard, causes a classifier, except the at least one classifier, that is included in the classifier set to exercise additional learning before the classifier set executes a classification process on goods that belong to the next group.SELECTED DRAWING: Figure 2 【課題】物品の分類に用いる分類器の分類精度が低下する前に該分類器の追加学習を行うことが可能な機械学習装置を実現する。【解決手段】複数のグループの各々に属する物品の状態を、該物品を被写体として含む画像を参照して分類する複数の分類器からなる分類器群（１０１）に、機械学習を行わせる機械学習装置１は、分類器群に含まれる各分類器の分類精度をグループ毎に評価する評価部（１０２）と、分類器群に含まれる少なくとも１つの分類器の分類精度が予め定められた基準に達しない場合、分類器群が次のグループに属する物品に対する分類処理を実行する前に、分類器群に含まれる、少なくとも１つの分類器以外の分類器に追加学習を行わせる学習制御部（１０３）と、を備える。【選択図】図２