Fast optimization classification algorithm based on ELM and SVM

Abstract

The invention discloses a fast optimization classification algorithm based on ELM and SVM. According to the fast optimization classification algorithm based on the ELM and the SVM, the linear non-separable problem is converted into the linear separable problem, and two SVM classifiers are trained as the algorithm core. This algorithm can greatly improve the network generalization performance and has high efficiency and can process more data in a short period of time so as to further promote wider application of the ELM and the SVM in pattern recognition, machine learning, big data processing and other fields.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a fast optimization classification algorithm based on ELM and SVM. Background technique [0002] In the field of machine learning, the problem of pattern classification is one of the most popular topics, and it is the key basic topic to solve the application problems of pattern recognition. Since the 1950s and 1960s, there have been many research results on the topic of pattern classification, among which the more representative basic learning algorithms are: perceptron learning algorithm, BP learning algorithm, SVM learning algorithm, and ELM learning algorithm, etc. However, compared with the application requirements, the current learning algorithms still have a considerable distance from meeting the application requirements in terms of learning effect and learning efficiency. [0003] (1) Extreme Learning Machine (ELM) is an emerging machine learning algorithm...

AI Technical Summary

Problems solved by technology

[0006] The technical advantages of SVM are concentrated in solving linearly separable classification problems, especially linearly separable binary classification problems. When dealing with nonlinear classification problems, the performance of SVM largely depends on the selection of kernel functions, but how There is no general solution for choosing a kernel function, which makes SVM hindered in engineering applications

In addition, after the kernel function is determined, SVM needs to solve the quadratic programming of the function when solving, which requires a lot of storage overhead and time overhead, which reduces its learning efficiency

Images