A Distance-Based Weighted Undersampling Scheme for Support Vector Machines and its Application to Imbalanced Classification

Abstract

A support vector machine (SVM) plays a prominent role in classic machine learning, especially classification and regression. Through its structural risk minimization, it has enjoyed a good reputation in effectively reducing overfitting, avoiding dimensional disaster, and not falling into local minima. Nevertheless, existing SVMs do not perform well when facing class imbalance and large-scale samples. Undersampling is a plausible alternative to solve imbalanced problems in some way, but suffers from soaring computational complexity and reduced accuracy because of its enormous iterations and random sampling process. To improve their classification performance in dealing with data imbalance problems, this work proposes a weighted undersampling (WU) scheme for SVM based on space geometry distance, and thus produces an improved algorithm named WU-SVM. In WU-SVM, majority samples are grouped into some subregions (SRs) and assigned different weights according to their Euclidean distance to the hyper plane. The samples in an SR with higher weight have more chance to be sampled and put to use in each learning iteration, so as to retain the data distribution information of original data sets as much as possible. Comprehensive experiments are performed to test WU-SVM via 21 binary-class and six multiclass publically available data sets. The results show that it well outperforms the state-of-the-art methods in terms of three popular metrics for imbalanced classification, i.e., area under the curve, F-Measure, and G-Mean.

IP102: A Large-Scale Benchmark Dataset for Insect Pest Recognition

“…The G-mean (denoted as GM) evaluates class-wise sensitivity and indicates the balanced classification performances on the majority and minority classes. The micro average scheme M AUC [15] is defined as the area under the curve metric. As for the task of object detection, we utilize the Average Precision (AP) (IoU=[.50:.05:.95]), AP .50 (IoU=.50), and AP .75 (IoU=.75) as performance evaluation metrics.…”

Self-Paced Balance Learning for Clinical Skin Disease Recognition

“…1) Sampling-Based Methods: Sampling-based methods attempt to handle the class imbalance problem at the data level, i.e., improving the data preprocessing technique. Specifically, these methods aim to balance the distribution of the original training set by over-sampling the minority classes [40]- [43], under-sampling the majority classes [7], [44], [45], or both.…”

“…As for the area under the curve (AUC) metric in the classification problem, we follow the micro average scheme M AUC of the definition as in [7]. Similar to the form of F -measure and G-mean, it integrates the weighted average of all labels:…”

“…However, most existing works on class imbalance problems focus only on the imbalanced distribution of sample numbers among different classes [7]- [9]. Such distribution indicates a large gap in the training numbers among categories [10]- [13], where there mainly exists three types of solutions, i.e., the sampling-based [14]- [16], the cost-sensitive based [17]- [19], and the ensemble-based methods [20], [20], [21].…”

A New Method of Fuzzy Support Vector Machine Algorithm for Intrusion Detection

“…The algorithms based on frequency can be realized at a lower computation cost by reducing the dimension of the frequency vectors [56]. SVM is sensitive to noises and outliers [57][58][59][60]. FSVM is based on fuzzy theory to reduce the influence of noises or outliers on the classification hyperplane [61][62][63][64][65][66].…”