Cost-Sensitive Broad Learning System for Imbalanced Classification and Its Medical Application

Liang Yao,Ziwen Wang,Xiaozheng Wang,Long Lei,Baoliang Zhao,Ying Hu,Pak Kin Wong

doi:10.3390/math10050829

Abstract

As an effective and efficient discriminative learning method, the broad learning system (BLS) has received increasing attention due to its outstanding performance without large computational resources. The standard BLS is derived under the minimum mean square error (MMSE) criterion, while MMSE is with poor performance when dealing with imbalanced data. However, imbalanced data are widely encountered in real-world applications. To address this issue, a novel cost-sensitive BLS algorithm (CS-BLS) is proposed. In the CS-BLS, many variations can be adopted, and CS-BLS with weighted cross-entropy is analyzed in this paper. Weighted penalty factors are used in CS-BLS to constrain the contribution of each sample in different classes. The samples in minor classes are allocated higher weights to increase their contributions. Four different weight calculation methods are adopted to the CS-BLS, and thus, four CS-BLS methods are proposed: Log-CS-BLS, Lin-CS-BLS, Sqr-CS-BLS, and EN-CS-BLS. Experiments based on artificially imbalanced datasets of MNIST and small NORB are firstly conducted and compared with the standard BLS. The results show that the proposed CS-BLS methods have better generalization and robustness than the standard BLS. Then, experiments on a real ultrasound breast image dataset are conducted, and the results demonstrate that the proposed CS-BLS methods are effective in actual medical diagnosis.

Highlights

The broad learning system (BLS) is an efficient and effective machine learning technique, which is designed by the inspiration of the random vector functional-link neural network (RVFLNN) [1,2]
Since the broad structure is significant to the performance of the BLS, a series of experiments are conducted on different broad structures to verify the performance of the cost-sensitive BLS algorithm (CS-BLS) methods and the standard BLS
The weighted penalty factor in the CS-BLS is utilized to provide a proper weight to each sample and constrain its contribution

Summary

Introduction

The broad learning system (BLS) is an efficient and effective machine learning technique, which is designed by the inspiration of the random vector functional-link neural network (RVFLNN) [1,2]. As the pseudo-inverse algorithm is used to compute the output weights of the standard BLS, the BLS has the characteristic of an efficient operation speed, and it is adopted in many real-world applications, such as medical data analysis [3], fault diagnosis [4], and robotics [5]. A gradient descent-based BLS was proposed in [7] for the control of nonlinear dynamic systems, which adopted gradient descent other than the pseudo-inverse algorithm to calculate the weight matrix in BLS iteratively. The imbalanced classification problem that suffers from imbalanced class distributions is encountered in many real-world domains, such as medical diagnosis [8], abnormal activity recognition [9], fault diagnosis [10], and fraud detection [11]. Taking breast cancer diagnosis as an example [12], detecting the minority class (i.e., the class of malignant lesions) should draw more attention, and the accurate diagnosis of it at early stages would increase the survival rate of patients

Methods

Results

Discussion

Conclusion