CNN-based width learning classification method

Abstract

The invention discloses a CNN-based width learning classification method. The method comprises the following specific steps of obtaining training data and test data; preprocessing the training data and the test data; performing feature extraction on the training data by using a convolutional neural network CNN to obtain feature mapping of the training data, and generating a feature node layer of the width learning basic model; enhancing the mapped features into an enhanced matrix with randomly generated weight, and generating an enhanced node layer of a width learning basic model; constructingan input matrix by the feature node layer and the enhanced node layer, inputting the input matrix into a width learning model for training, and constructing a width learning basic model; and utilizing the finally trained improved width learning model. According to the technical scheme, under the condition that it is guaranteed that the number of feature nodes is similar, CNN_BLS has a better classification result than BLS, meanwhile, in comparison with an ELM model, the original high efficiency of a width learning system is reserved, and a better comprehensive effect is achieved.

Description

technical field [0001] The invention relates to the field of intelligent learning, in particular to a CNN-based width learning classification method. Background technique [0002] At present, in the field of intelligent learning, deep neural network models are widely used to solve different types of intelligent classification problems. However, when faced with more complex problems, it is generally necessary to increase the depth of the network structure of the deep neural network model and adjust the number of neurons in each layer of the network, and then use an iterative update method to train the connection weights between each network layer. Finally, the ideal model effect is achieved. In the face of a large amount of experimental data, as the depth model becomes more and more complex and the number of layers becomes deeper, the parameters to be optimized also increase exponentially. It usually takes a lot of time and machine resources to iteratively optimize the param...

AI Technical Summary

Problems solved by technology

[0004] However, since BLS uses randomly generated weights and biases by default in the process of solving the description characteristics of samples and constructing feature layers, in the face of complex samples, it will cause insufficient description of sample representative features, thereby reducing learning ability. problem, the accuracy of the classification task is also lost

Images