Image classification method based on cortical thalamus computing model

Abstract

The invention relates to the field of brain-like intelligence and artificial intelligence, specifically relates to an image classification method based on a cortical thalamus computing model and aimsto solve problems that training data is wasted in the traditional artificial neural network and a large amount of calculation is required in the process of training the neural network. According to the invention, classification mark prediction is performed on an input image respectively based on a contour prior neural network N1 and a neural network N2 which integrates a thalamus regulation and control effect, and two prediction results are integrated according to a preset weight to obtain the classification of the input image. The image classification method performs an image classification test by using an MNIST data set and a FashionMNIST data set under a small sample data training condition, and the test result shows that the performance of the image classification method based on thecortical thalamus computing model is more excellent than that of the traditional artificial neural network.

Description

technical field [0001] The invention relates to the fields of brain-like intelligence and artificial intelligence, in particular to an image classification method based on a cortical and thalamic computing model. Background technique [0002] In primates, the thalamus receives sensory information from all over the body, except smell, and projects it to the cerebral cortex. In addition to serving as a sensory information conversion station, the thalamus also plays a regulatory role in the process of cortical information transmission. The thalamus is mainly composed of lateral geniculate body (LGN), thalamic reticular nucleus (TRN) and pulvinar. The lateral geniculate body connects the sensory organs with the cerebral cortex, and plays the role of information transfer; the reticular nucleus of the thalamus regulates the information transmission between the cerebral cortex and the thalamus; the pulvinus of the thalamus is the largest nucleus in the thalamus, accounting for nea...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problem of wasting training data in the traditional artificial neural network and requiring a large amount of calculation in the process of training the neural network, the present invention proposes an image classification based on the cortex-thalamus computing model method, the method is based on the contour prior neural network N 1 , A neural network that integrates the regulation of the thalamus N 2 , predict the classification mark of the input image respectively, and fuse the two prediction results according to the preset weight to obtain the classification of the input image

Images