A soft-interval support vector machine classification method based on neurodynamics

Abstract

The invention discloses a soft-interval support vector machine classification method based on neurodynamics, which comprises the following steps: 1) introducing a training set and generating a soft-interval support vector machine coefficient matrix; 2) designing a convex quadratic programming problem constrained by equality and inequality by using the coefficient matrix in step 1); 3) equivalentlyconverting the convex quadratic programming problem constrained by equality and inequality in the step 2) into a linear variational inequality; 4) equivalently converting the linear variational inequality in the step 3) into a piecewise linear projection equation; 5) obtaining an optimal solution of the piecewise linear projection equation in the step 4) by using a neurodynamic method, and obtaining a decision function of a soft-interval support vector machine; 6), import that test set and use the decision function in step 5) to classify. The invention adopts the neurodynamic solver to solve,and has the advantages of high sensitivity, better specificity, stronger real-time performance and higher accuracy.

Description

technical field [0001] The invention relates to the field of soft interval support vector machine classification methods, in particular to a neural dynamics-based soft interval support vector machine classification method. Background technique [0002] Among the various methods of pattern recognition, support vector machine is an important classification technique in machine learning, which is widely used in handwritten digit recognition, object recognition, face detection, text classification and speech recognition. The main goal of SVM is to construct an optimal decision function (hyperplane) that can distinguish data points given a set of sample features. The support vector machine classifier mainly divides the data points represented by an n-dimensional vector into different classes by constructing a hyperplane. In addition, SVM selects the optimal hyperplane with the largest margin to achieve the largest separation of data points. [0003] The basic support vector mac...

AI Technical Summary

Problems solved by technology

These numerical algorithm solvers usually have the advantage that they can significantly increase the calculation speed, however, due to the increase in the number of iterations, the numerical algorithm solver's operation accuracy may decrease

Also, in large-scale real-time applications, numerical algorithm solvers may degrade performance due to their serial processing nature

Images