Defect detection model interpretation method and device, equipment and storage medium

• Main Authors: WANG WEI, ZHU ZHILING, GUO XINGCHENG, FAN QIQIANG
• Format: Patent
• Language: Chinese; English
• Published: 30.05.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a defect detection model interpretation method and device, equipment and a storage medium. The method comprises the steps that a deep convolutional neural network is established, and a defect detection model and the probability that each metric element has defects are obtained through training; performing knowledge distillation processing on the defect detection model to obtain a soft decision tree corresponding to the defect detection model; obtaining a target data set, carrying out an addition operation on each metric unit instance in the target data set, inputting the data after the addition into the soft decision tree, and obtaining the training loss of the soft decision tree; comparing the training loss of training after each metric element is added with the loss of the defect detection model, and making a difference between the training loss and the loss of the defect detection model; and determining the sensitivity of each metric unit according to a result of comparing and subtr