Target object influence factor screening method and device, equipment and storage medium

Abstract

The invention relates to a data analysis technology, and discloses a target object influence factor screening method, which comprises the following steps: inputting a standard image into a classification model to obtain a category prediction value of which the standard image is a target category; calculating the category prediction value to obtain the number of influence factors to be selected; calculating the plurality of influence factors by using a classification model to obtain an influence prediction value of each influence factor, and selecting the quantity of influence factors from a plurality of preset influence factors as a target factor set according to the influence prediction values; and calculating a label value of each influence factor in the target factor set, and selectingthe influence factor of which the label value is a preset label threshold value from the target factor set as a standard influence factor. The invention further provides a target object influence factor screening device and a computer readable storage medium. In addition, the invention also relates to a blockchain technology, and the standard image can be stored in the blockchain node. According to the invention, the influence factors in the input data of the deep learning model can be obtained.

Description

technical field [0001] The present invention relates to the technical field of data analysis, in particular to a screening method, device, electronic equipment and computer-readable storage medium for influencing factors of targets. Background technique [0002] Interpretability of deep learning models for images has long been a key concern in machine learning. Specifically, the interpretability of the deep learning model plays an important role in the improvement of the model, the credibility of the model, and the transparency of the model. [0003] Feature visualization is currently a commonly used method of explaining deep learning models for images. In the process of feature visualization, we can find the influencing factors in the input data of the model, and visualize the influence of each influencing factor on the prediction results of the model, which can explain the model to a certain extent. The focus on the input data in making predictions. However, in practical...

AI Technical Summary

Problems solved by technology

However, in practical applications, feature visualization can only show the scope of the influencing factors in the input data that the model pays attention to when making data label predictions for the input data (such as position coordinates), but it cannot reflect the specific model based on the input data. Which specific influencing factors of the

For example, when generating predicted labels for image recognition, feature visualization can only show the pixel positions of the influencing factors that the model focuses on in the image, but cannot reflect the label predictions made by the model based on specific influencing factors in the image

Images