Solve Class Imbalance Problem Research Articles

Borderline-DEMNET: A Workflow for Detecting Alzheimer’s and Dementia Stage by Solving Class Imbalance Problem

Alzheimer’s Disease (AD) is the leading cause of dementia, a broad term encompassing memory loss and other cognitive impairments. Although there is no known cure for dementia, managing specific symptoms associated with it can be effective. Mild dementia stages, including AD, can be treated, and computer-based techniques have been developed to aid in early diagnosis. This paper presents a new workflow called Borderline-DEMNET, designed to classify various stages of Alzheimer’s/dementia with more than three classes. Borderline-SMOTE is employed to address the issue of imbalanced datasets. A comparison is made between the proposed Borderline-DEMNET workflow and the existing DEMNET model, which focuses on classifying different dementia and AD stages. The evaluation metrics specified in the paper are used to assess the results. The framework is trained, tested, and validated using the Kaggle dataset, while the robustness of the work is checked using the ADNI dataset. The proposed workflow achieves an accuracy of 99.17% for the Kaggle dataset and 99.14% for the ADNI dataset. In conclusion, the proposed workflow outperforms previously identified models, particularly in terms of accuracy. It also proves that selecting a proper class balancing technique will increase accuracy.

Borderline-DEMNET: A Workflow for Detecting Alzheimer’s and Dementia Stage by Solving Class Imbalance Problem

Alzheimer’s Disease (AD) is the leading cause of dementia, a broad term encompassing memory loss and other cognitive impairments. Although there is no known cure for dementia, managing specific symptoms associated with it can be effective. Mild dementia stages, including AD, can be treated, and computer-based techniques have been developed to aid in early diagnosis. This paper presents a new workflow called Borderline-DEMNET, designed to classify various stages of Alzheimer’s/dementia with more than three classes. Borderline-SMOTE is employed to address the issue of imbalanced datasets. A comparison is made between the proposed Borderline-DEMNET workflow and the existing DEMNET model, which focuses on classifying different dementia and AD stages. The evaluation metrics specified in the paper are used to assess the results. The framework is trained, tested, and validated using the Kaggle dataset, while the robustness of the work is checked using the ADNI dataset. The proposed workflow achieves an accuracy of 99.17% for the Kaggle dataset and 99.14% for the ADNI dataset. In conclusion, the proposed workflow outperforms previously identified models, particularly in terms of accuracy. It also proves that selecting a proper class balancing technique will increase accuracy.

A comprehensive analysis of feature ranking-based fish disease recognition

In recent years, the field of emerging computer vision systems has witnessed significant advancements in automated disease diagnosis through the utilization of vision-oriented technology. This article proposes an optimal approach for detecting the presence of ailments in Rohu fish. The aims of our research is to identify the most significant features based on Analysis of Variance (ANOVA) feature selection and evaluate the best performance among all features for Rohu fish disease recognition. At the outset, diverse techniques for image preprocessing were employed on the acquired images. The region affected by the disease was partitioned through utilization of the K-means clustering algorithm. Subsequently, 10 distinct statistical and Gray-Level Co-occurrence Matrix (GLCM) features were extracted after the image segmentation. The ANOVA feature selection technique was employed to prioritize the most significant features N (where 5 ≤ N ≤ 10) from the pool of 10 categories. The Synthetic Minority Oversampling Technique, often known as SMOTE, was applied to solve class imbalance problem. After conducting training and testing on nine different machine learning (ML) classifiers, an evaluation was performed to estimate the performance of each classifier using eight various performance metrics. Additionally, a receiver operating characteristic (ROC) curve was generated. The classifier that utilized the Enable Hist Gradient Boosting algorithm and selected the top 9 features demonstrated superior performance compared to the other eight models, achieving the highest accuracy rate of 88.81%. In conclusion, we have demonstrated that the feature selection process reduces the computational cost.

A novel SMOTE-based resampling technique trough noise detection and the boosting procedure

Most of the classification methods assume that the numbers of class observations are balanced. In such cases, models are predicted by giving biased weight to the the class with more observations. Therefore, the classifiers ignore the class with smaller number of observations and the majority class makes biased predictions. There are some advised performance measures to be used in datasets, as well as recommended approaches to solve class imbalance problem. One of the most widely used methods is resampling method. In this study, the difficulties relevant to random oversampling (ROS) and synthetic minority oversampling technique (SMOTE), which are some of the oversampling methods, are discussed. This study aims to propose a combination of a new noise detection method and SMOTE to overcome those difficulties. Using the boosting procedure in ensemble algorithms, noise detection is possible with the proposed SMOTE with boosting (SMOTEWB) method, which makes use of this information to determine the appropriate number of neighbors for each observation within SMOTE algorithm.

Learning automata based particle swarm optimization for solving class imbalance problem

Class imbalance is an important problem in many domains such as disease classification, network intrusion detection, fraud detection, and spam filtering. While dealing with imbalanced datasets, traditional supervised machine learning algorithms do not often provide acceptable results. Several approaches are used to handle the class imbalance problem. Of these, undersampling approaches are mostly followed by the researchers in which the number of instances in the majority class gets reduced. Selection of instances from the majority class can be considered as an optimization problem. To this end, in this paper, we present an undersampling approach based on widely-used Particle Swarm Optimization (PSO). The majority class samples are first clustered to form the initial undersampled set. The samples to be selected are then optimized using PSO to give the best model. The parameters of PSO are fine tuned using Learning Automata. Appropriate metrics suitable for class imbalance problems have been used to construct the fitness function for optimizing the undersampled training set. The proposed method has achieved 2% to 10% performance improvement over most of the contemporary methods on various datasets with imbalance ratios ranging from 5 to 130, thus showing that the method is robust and useful in practical scenarios. The code of the proposed method can be accessed via https://github.com/kkg1999/Undersampling.

A Novel Hybrid Sampling Algorithm for Solving Class Imbalance Problem in Big Data

The uneven distribution of classes in any dataset poses a tendency of biasness toward the majority class when analyzed using any standard classifier. The instances of the significant class being deficient in numbers are generally ignored and their correct classification which is of paramount interest is often overlooked in calculating overall accuracy. Therefore, the conventional machine learning approaches are rigorously refined to address this class imbalance problem. This challenge of imbalanced classes is more prevalent in big data scenario due to its high volume. This study deals with acknowledging a sampling solution based on cluster computing in handling class imbalance problems in the case of big data. The newly proposed approach hybrid sampling algorithm (HSA) is assessed using three popular classification algorithms namely, support vector machine, decision tree and k-nearest neighbor based on balanced accuracy and elapsed time. The results obtained from the experiment are considered promising with an efficiency gain of 42% in comparison to the traditional sampling solution synthetic minority oversampling technique (SMOTE). This work proves the effectiveness of the distribution and clustering principle in imbalanced big data scenarios.

Hybridization of ring theory-based evolutionary algorithm and particle swarm optimization to solve class imbalance problem

Many real-life datasets are imbalanced in nature, which implies that the number of samples present in one class (minority class) is exceptionally less compared to the number of samples found in the other class (majority class). Hence, if we directly fit these datasets to a standard classifier for training, then it often overlooks the minority class samples while estimating class separating hyperplane(s) and as a result of that it missclassifies the minority class samples. To solve this problem, over the years, many researchers have followed different approaches. However the selection of the true representative samples from the majority class is still considered as an open research problem. A better solution for this problem would be helpful in many applications like fraud detection, disease prediction and text classification. Also, the recent studies show that it needs not only analyzing disproportion between classes, but also other difficulties rooted in the nature of different data and thereby it needs more flexible, self-adaptable, computationally efficient and real-time method for selection of majority class samples without loosing much of important data from it. Keeping this fact in mind, we have proposed a hybrid model constituting Particle Swarm Optimization (PSO), a popular swarm intelligence-based meta-heuristic algorithm, and Ring Theory (RT)-based Evolutionary Algorithm (RTEA), a recently proposed physics-based meta-heuristic algorithm. We have named the algorithm as RT-based PSO or in short RTPSO. RTPSO can select the most representative samples from the majority class as it takes advantage of the efficient exploration and the exploitation phases of its parent algorithms for strengthening the search process. We have used AdaBoost classifier to observe the final classification results of our model. The effectiveness of our proposed method has been evaluated on 15 standard real-life datasets having low to extreme imbalance ratio. The performance of the RTPSO has been compared with PSO, RTEA and other standard undersampling methods. The obtained results demonstrate the superiority of RTPSO over state-of-the-art class imbalance problem-solvers considered here for comparison. The source code of this work is available in https://github.com/Sayansurya/RTPSO_Class_imbalance.

Improved accuracy of pest detection using augmentation approach with Faster R-CNN

In agriculture, Pests are decreasing agricultural productivity. Identifying a pest is a challenging process and subject to expert opinion. Nowadays, lots of work carried out for automatic pest detection. It becomes possible because of emerging Deep Learning’s object detection architectures. This paper shows the multi-class pest detection using Faster R-CNN architecture and compared the performance results of image augmentation with focused on the accuracy performance along with small dataset. We have used Horizontal Flip and 90 Degree Rotation augmentation parameters for solving class imbalance problem. We found that trained pest detection model with augmentation options can perform better with an accuracy of 91.02% using Faster R-CNN architecture.

Synthetic oversampling based decision support framework to solve class imbalance problem in smoking cessation program

ABSTRACT Smoking is one of the significant avoidable risk factors for premature death. Most smokers make multiple quit attempts during their lifetime but smoking dependence is not easy and many people eventually failed quit attempts. Predicting the likelihood of success in smoking cessation program is necessary for public health. In recent years, a few numbers of decision support systems have been developed for dealing with smoking cessation based on machine learning techniques. However, the class imbalance problem is increasingly recognized as serious in real-world applications. Therefore, this paper presents a synthetic minority over-sampling technique (SMOTE) based decision support framework in order to predict the success of smoking cessation program using Korea National Health and Nutrition Examination Survey (KNHANES) dataset. We carried out experiments as follows: I) the unnecessary instances and variables have been eliminated, II) then we employed three variations of SMOTE, III) also the prediction models have been constructed. Finally, compare the prediction models to obtain the best model. Our experimental results showed that SMOTE improved the prediction performance of machine learning classifiers among evaluation metrics. Moreover, SMOTE regular based Random Forest (RF) and Naive Bayes (NB) classifiers were determined the best prediction models in real-world smoking cessation dataset. Consequently, our decision support framework can interpret the important risk factors of smoking cessation using multivariate regression analysis.

Ss-InfoGAN for Class-Imbalance Classification of Bearing Faults

As the core part of the Prognostic and Health Management (PHM) of major equipment such as high-speed trains and aero engines, bearing fault classification have been the research priorities in the field. Although convolutional neural network (CNN) has shown good results in this type of task, the real application with limited training data makes CNN have a big gap between the actual application and the expected effect. Therefore, bearing faults classification with class-imbalance is a very practical work. In this paper, semi-supervised information maximizing generative adversarial network (ss-InfoGAN), which uses adversarial structure to generate samples of the minority, is introduced to augment data to solve class imbalance problem. In addition, the latent codes, the inputs of generator, are decomposed into three parts with three additional networks, respectively, at the start of generator. Meanwhile, the 50% precision threshold is proposed during the training stage of discriminator to make a trade-off between computing resources and theoretical foundations and facilitate the network converge. Bearing fault experiments are conducted to investigate the effectiveness of the presented network. The result shows classification accuracy is improved by 40% by the ss-InfoGAN compared to the traditional CNN for the case of extremely class-imbalance condition.

An Efficient Domain-Adaptation Method using GAN for Fraud Detection

In this paper, an efficient domain-adaptation method is proposed for fraud detection. The proposed method employs the discriminative characteristics used in feature maps and generative adversarial networks (GANs), to minimize the deviation that occurs when a common feature is shifted between two domains. To solve class imbalance problem and increase the model’s detection accuracy, new data samples are generated by applying a minority class data augmentation method, which uses a GAN. We evaluate the classification performance of the proposed domain-adaption model by comparing it against support vector machine (SVM) and convolutional neural network (CNN) models, using classification performance evaluation indicators. The experimental results indicated that the proposed model is applicable to both test datasets; furthermore, it requires less time for learning. Although the SVM offers a better detection performance than the CNN and proposed domain-adaptation model, its learning time exceeds those of the other two models when dataset increases. Also, although the detection performance of the CNN-based model is similar to that of the proposed domain-adaptation model, its learning process is longer. In addition, although the GAN used to solve the class imbalance problem of the two datasets requires slightly more time than SMOTE (synthetic minority oversampling technique), it shows a better classification performance and is effective for datasets featuring class imbalances.

Influencing web customers. Deep learning practical application

Abstract With the global expansion of internet businesses, companies look for possibilities to increase their turnovers in virtual media. In these circumstances, it is significant for companies to employ online user behavior data in order to understand user intents and, to some extent, influence user online activity. This research examines the practical solution of the stated goal through an application of deep learning models. The contribution of the current study is an investigation of the success of real-time influence on web portal users to conduct specific activities leading to the achievement of predefined business goals. The study investigates several top-rated over-sampling techniques while solving class imbalance problem to evaluate their impact on the deep learning model performance. The study was conducted on the Lithuanian online educational platform. The obtained results indicate the feasibility of the suggested method.

Investigating factors affecting customer churn in electronic banking and developing solutions for retention

Due to the competitive environment in electronic banking services, banks need to understand situations that lead to customer churn. Hence identification of factors affecting customer churn and developing programs to retain customer is important for banks. In this study, binomial logistic regression technique is used to identify factors affecting customer churn in electronic banking. Then affecting factors are employed in decision tree and artificial neural network methods to predict customer churn. Bagging technique is used to solve class imbalance problem and improves accuracy. The results show that the length of customer association, customer's age, customer's gender and the number of mobile banking transactions influence customer churn. Comparison of the prediction methods shows that performance of decision tree classifier is better than artificial neural network. Finally, based on the findings, some solutions are recommended to prevent customer from churning. In addition, some methods are suggested to build customer churn prediction model.

Bug Severity Prediction using Class Imbalance Problem

Class imbalance problem is often observed when instances of major class exceed instances of minor class. The performance of machine learning techniques is immensely afflicted by imbalanced data in several fields. The skewed distribution either predicts the majority class with high error rate or will not foresee the minority class. To solve the problem of imbalanced data of software bugs, Synthetic minority oversampling technique (SMOTE) is used which balances the imbalanced datasets of Apache Projects. It is applied on summary of bugs to balance the dataset and predicts severity at system and component level. Several machine learning techniques are applied on imbalanced as well as balanced datasets to predict the severity of software bugs using textual description. Test outcomes and statistical analysis shows improved results on balanced datasets in respect to Gmean and balance metrics instead of machine learning techniques applied on imbalanced data. Evaluation metrics Gmean improves by 34% and balance by 11% at system level and by 42% and 62% at component level. Further, it was observed that solving class imbalance problem on textual data is helpful in augmenting the performance.

Performance Analysis of Under-Sampling and Over-Sampling Techniques for Solving Class Imbalance Problem

Most of the traditional classification algorithms assume their training data to be well-balanced in terms of class distribution. Real-world datasets, however, are imbalanced in nature thus degrade the performance of the traditional classifiers. An imbalance data-set typically make prediction accuracy difficult. Data pre-processing approaches discuss this issue by using random under-sampling or oversampling techniques. To solve this problem, many strategies are adopted to balance the class distribution at the data level. The data level methods balance the imbalance distribution between majority and minority classes using either oversampling or under-sampling techniques. In this paper, we present the performance analysis of under-sampling method and oversampling methods. The methods are implemented with 5 conventional classifiers like C4.5 Decision Tree (DT), k-Nearest Neighbor (k-NN), Multilayer Perceptron (MLP), Support Vector Machine (SVM), and Naive Bayes (NB) on 15 real life data sets. The experimental results show comparative study of under-sampling and over sampling technique.

Solving class imbalance problem using bagging, boosting techniques, with and without using noise filtering method

In numerous real-world applications/domains, the class imbalance problem is prevalent/hot topic to focus. In various existing work, for solving class imbalance problem, almost data is labeled as one class called majority class, while fewer data is la

Issues and challenges of class imbalance problem in classification

Class imbalance problem is the problem of classification when we seek out exceptional cases using traditional classification algorithms. Traditional classification algorithms are designed to look for either bigger classes or classes with the similar size. These algorithms when used to identify smaller class from the data either fails to detect or gives erroneous results. Researchers have worked on this problem using various concepts, logics or by modifying existing classification algorithms. This paper discusses existing research trends used to solve class imbalance problem. It also highlights the issues and gaps related to this problem.

Improving classification of mature microRNA by solving class imbalance problem.

MicroRNAs (miRNAs) are ~20–25 nucleotides non-coding RNAs, which regulated gene expression in the post-transcriptional level. The accurate rate of identifying the start sit of mature miRNA from a given pre-miRNA remains lower. It is noting that the mature miRNA prediction is a class-imbalanced problem which also leads to the unsatisfactory performance of these methods. We improved the prediction accuracy of classifier using balanced datasets and presented MatFind which is used for identifying 5′ mature miRNAs candidates from their pre-miRNA based on ensemble SVM classifiers with idea of adaboost. Firstly, the balanced-dataset was extract based on K-nearest neighbor algorithm. Secondly, the multiple SVM classifiers were trained in orderly using the balance datasets base on represented features. At last, all SVM classifiers were combined together to form the ensemble classifier. Our results on independent testing dataset show that the proposed method is more efficient than one without treating class imbalance problem. Moreover, MatFind achieves much higher classification accuracy than other three approaches. The ensemble SVM classifiers and balanced-datasets can solve the class-imbalanced problem, as well as improve performance of classifier for mature miRNA identification. MatFind is an accurate and fast method for 5′ mature miRNA identification.

Resampling Methods for Solving Class Imbalance Problem in Traffic Incident Detection

Traffic data is highly skewed with rare traffic incidents in the real word while most of the existing automatic incident detection (AID) algorithms suffer from limitations due to their inability to detect incidents under imbalanced traffic data condition. This paper developed feasible AID algorithms based on resampling methods to process imbalanced traffic data. In order to obtain the optimal sampling method for incident detection, we compare the detection performance of different AID algorithms based on various resampling methods. The detection performance is evaluated by the common criteria including classification rate, detection rate, false alarm rate, mean time to detection and an integrated performance index. The I-880 dataset is finally used in experiments to verify the proposed algorithms. The experimental results indicate that the proposed AID algorithm based on resampling can achieve better performance through handling imbalanced traffic data problem. Moreover, the under-sampling is competitive than over-sampling for traffic incident detection.

Adapting rough-fuzzy classifier to solve class imbalance problem in heart disease prediction using FCM

The main objective of this research is to develop a heart disease prediction technique by solving class imbalance problem. Class imbalance problem severely affects the performance of the prediction if the distribution of data is not clearly defined. To overcome class imbalance problem and achieve promising results in this work, the proposed technique is divided into three steps. Initially, the input data is given to fuzzy c–means clustering algorithm that converts the original data into equal number samples for all the classes. Then, rules are generated from the rough set theory and these rules are used for prediction with the fuzzy classifier. For testing, test data is converted into relevant space after matching with the original cluster centres and then, sample is tested with rough–fuzzy classifier. The results prove that the proposed technique generated excellent results by achieving the accuracy of 81% in Cleveland and 80% in Hungarian datasets.