OIN (Optimal Input Normalization) neural network training method for mixed SVM (Support Vector Machine) regression algorithm

Abstract

The invention discloses an OIN (Optimal Input Normalization) neural network training method for a mixed SVM (Support Vector Machine) regression algorithm. The method mainly comprises the steps of an OIN forward propagation part, a SVM regression part and an OIN backward propagation part; and finally, an optimized OIN / SVM mixed model is obtained through various trainings; and in a testing stage, a testing sample is input into the optimized OIN / SVM mixed model to obtain a predicated result, so that the classification of the sample or the regression of the sample is predicated. According to the invention, a latest designed OIN artificial neural network training method is adopted; and through the adoption of the method disclosed by the invention, the convergence of the traditional backward propagation algorithm can be greatly improved.

Description

technical field [0001] The invention relates to a data analysis method, in particular to an OIN neural network training method of a mixed SVM regression algorithm. Background technique [0002] Artificial Neural Network (ANN, Artificial Neural Network) is an optimized machine learning method and is widely used in various data analysis fields to replace or supplement polynomial-based regression analysis and classification. Existing neural network applications can only be limited to simple designs with a small number of design parameters, and the required data set size for modeling should increase geometrically or exponentially with the number of design parameters. Therefore, a neuron analysis requires There are a large number of sufficient density distributions and experimental data, on the other hand, the computational cost increases accordingly, and the inefficient use of a large amount of data in the design space may also lead to excessive waste of computational cost. [...

AI Technical Summary

Problems solved by technology

However, the output described according to the mechanism is a Boolean category value directly obtained by the SVM classification algorithm, and the Boolean value is difficult to apply to the backpropagation process of the neural network; therefore, the mechanism does not specifically point out ANN-SVM How to train and how to implement weight correction

[0008] The present invention is suitable for solving related problems in data analysis. For example, the problem of color constancy in image segmentation, that is, the problem of image segmentation under the influence of different illuminations. The existing image segmentation does not use learning algorithms, but only uses the conversion of color space , under the influence of complex lighting, it cannot meet the ideal requirements

Images