Image Classification Using Biomimetic Pattern Recognition with Convolutional Neural Networks Features.

Liangji Zhou,Qingwu Li,Yan Zhou,Guanying Huo

doi:10.1155/2017/3792805

Abstract

As a typical deep-learning model, Convolutional Neural Networks (CNNs) can be exploited to automatically extract features from images using the hierarchical structure inspired by mammalian visual system. For image classification tasks, traditional CNN models employ the softmax function for classification. However, owing to the limited capacity of the softmax function, there are some shortcomings of traditional CNN models in image classification. To deal with this problem, a new method combining Biomimetic Pattern Recognition (BPR) with CNNs is proposed for image classification. BPR performs class recognition by a union of geometrical cover sets in a high-dimensional feature space and therefore can overcome some disadvantages of traditional pattern recognition. The proposed method is evaluated on three famous image classification benchmarks, that is, MNIST, AR, and CIFAR-10. The classification accuracies of the proposed method for the three datasets are 99.01%, 98.40%, and 87.11%, respectively, which are much higher in comparison with the other four methods in most cases.

Highlights

Image classification and recognition is a sophisticated task for machine, and it has been a hot issue in the field of Artificial Intelligence (AI) all the time
Convolutional Neural Networks (CNNs) ← X; % the raw training data are sent into CNN to get extracted feature vectors F = {F(1), F(2), . . . , F(K)}; % the extracted feature vectors are mapped into high-dimensional space to be % covered by Complex Geometry Coverage (CGC) class by class for i 1 to K do
It can be seen that Histogram of Oriented Gradient (HOG) and Biomimetic Pattern Recognition (BPR) perform much better than the other methods in the case of small-sized homogeneous datasets, while with the increase of training samples, CNN-SVM surpasses the Principal Component Analysis (PCA)-BPR, which means that CNN can better represent the feature than HOG and PCA do in the case of large-scale heterogeneous datasets

Summary

Introduction

Image classification and recognition is a sophisticated task for machine, and it has been a hot issue in the field of Artificial Intelligence (AI) all the time. A new method that combines CNNs with BPR is proposed to reduce the complexity of training networks and to improve the performance of classification. CNNs are used to automatically learn feature vectors from raw images, and the learned feature vectors are projected into highdimensional space to be covered by BPR classifier. Such a combination is expected to combine the advantages of CNNs on feature representation and BPR on classification.

Related Works

The Proposed Model

Experiments and Discussions

Conclusion