Deep learning characteristic visualization and model assessment method

Abstract

The invention discloses a deep learning characteristic visualization and model assessment method, and the method comprises the steps: inputting image data into a deep learning network model from an image database, carrying out the forwarding propagation of the inputted image data for one time through a convolution layer, a nonlinear layer and a pooling layer, and obtaining a classification result; calculating a characteristic pattern of a corresponding inputted image through a back propagation algorithm or a deconvolution algorithm or an LRP algorithm; carrying out the visualization analysis based on the characteristic pattern, and evaluating the deep learning network model.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to feature visualization and model evaluation methods of deep learning. Background technique [0002] Deep learning is a branch of machine learning that attempts to use algorithms that perform high-level abstraction on data using complex structures or multiple processing layers composed of multiple nonlinear transformations. [0003] Deep learning is a method based on representation learning of data in machine learning. [0004] Observations (such as an image) can be represented in a variety of ways, such as a vector of intensity values ​​for each pixel, or more abstractly as a series of edges, regions of a specific shape, etc. Whereas it is easier to learn tasks from examples (e.g., face recognition or facial expression recognition) using some specific representations. [0005] The advantage of deep learning is to use unsupervised or semi-supervised feature learni...

AI Technical Summary

Problems solved by technology

[0006] Although the existing deep learning research has achieved good results in image classification, speech recognition and other fields, due to its multi-layer nonlinear structure, deep learning is li

Images