Research And Application Of Classification Problem Based On Extreme Learning Machine

In recent years,artificial intelligence has developed rapidly.As the core content of artificial intelligence machine learning has been widely used in various fields of artificial intelligence: for example,in pattern recognition,natural language processing,computer vision,expert systems,intelligent robots,etc.Neural network is an important method of machine learning.Researchers have studied neural networks deeply,based on neural networks have put forward a variety of algorithms.The algorithm has been well applied in the regression and classification problems.Extreme Learning Machine(ELM)attracts the attention of many scholars because of its simple theory and easy implementation.However,extreme learning machine is still being with some problems.For example,irregular distribution of data,noise,and outliers in practical problems can seriously affect the classification accuracy of extreme learning machines.In supervised learning,the limited number of data samples will bring insufficient learning problem which affect the generalization ability of ELM algorithm.In this paper,we study the extreme learning machine mainly focus on the above problems.The main research results are as follows:In order to effectively extract the sensitive information of face image from high-dimensional image data and improve the face recognition effect,this paper proposes an extreme learning machine(ELM)with supervised sparse permutation based on manifold learning(SSLPP).SSLPP dynamically determines the local linearized neighborhood range of the face image data by calculating the neighborhood information,and accurately obtains the global and local discriminant information of the face image data.The discriminant information then improves the classification performance of the ELM classifier.Due to limited data samples in supervised learning,ELM algorithm faces problem of in sufficient learning in classification of hyperspectral remote sensing images.In order to overcome the above problems,this paper proposes a Locality Preserving Projection extreme learning machine(LPELM).The algorithm not only inherits the advantages of ELM but also considers the geometric structure and potential discriminating information between data points,thereby improving the generalization performance and classification accuracy of ELM.The experimental results show that these two optimized algorithms have achieved good classification effect when applied to face images and hyperspectral remote sensing images.