The Image Classification Method with CNN-XGBoost Model Based on Adaptive Particle Swarm Optimization

Wenjiang Jiao,Chao Qin,Xingwei Hao

doi:10.3390/info12040156

Abstract

CNN is particularly effective in extracting spatial features. However, the single-layer classifier constructed by activation function in CNN is easily interfered by image noise, resulting in reduced classification accuracy. To solve the problem, the advanced ensemble model XGBoost is used to overcome the deficiency of a single classifier to classify image features. To further distinguish the extracted image features, a CNN-XGBoost image classification model optimized by APSO is proposed, where APSO optimizes the hyper-parameters on the overall architecture to promote the fusion of the two-stage model. The model is mainly composed of two parts: feature extractor CNN, which is used to automatically extract spatial features from images; feature classifier XGBoost is applied to classify features extracted after convolution. In the process of parameter optimization, to overcome the shortcoming that traditional PSO algorithm easily falls into a local optimal, the improved APSO guide the particles to search for optimization in space by two different strategies, which improves the diversity of particle population and prevents the algorithm from becoming trapped in local optima. The results on the image set show that the proposed model gets better results in image classification. Moreover, the APSO-XGBoost model performs well on the credit data, which indicates that the model has a good ability of credit scoring.

Highlights

Image classification which belongs to the main research content of image processing has a broad application prospect in many sciences, such as object recognition, content understanding and image matching
Particle swarm optimization (PSO)-CNN-XGBoost represents the CNN-XGBoost model optimized by ordinary PSO, and the other two represent high-performance methods: CIDBM [34], PCAnet [35]
The reason is that the ensemble classifier XGBoost understands image features better than other classifiers, and it is more compatible with CNN due to stronger classification performance

Summary

Introduction

Image classification which belongs to the main research content of image processing has a broad application prospect in many sciences, such as object recognition, content understanding and image matching. Support vector machine (SVM) [1], k-nearest neighbor (KNN) [2] and decision tree (DT) [3] are all typical machine learning methods applied in this field These studies prove the effectiveness and reliability of machine learning applied in image classification. CNN is an efficient neural network learning model, whose convolution kernel in the convolutional layer plays an important role in the extraction of features. Kearns and Valiant [11] showed that weak classifiers can generate high precision estimates by integrating, as long as data is sufficient These studies proved that ensemble learning has a better learning ability than a single classifier. The experimental results on MNIST and CIFAR-10 show that the performance of this method is better than other methods, which verifies the effectiveness of the combination of CNN and XGBoost in image classification

Objectives

Methods

Results

Discussion

Conclusion