Data Imbalance Problem Research Articles

FDF-HybridFS: Towards design of a failure detection framework using hybrid feature selection method for IP core networks that connect 5G core in NFV-based test environment

With the advancement of recent technologies, high-speed and secure internet connectivity is required for 5th-generation mobile networks. A massive number of smart devices connected to the networks generate a large amount of data. Consequently, increasing traffic has generated more strain on networks and devices. Therefore, to handle the huge amount of data generated by the various devices and to detect network and device failures, automation is expected to play an important role. Hence, these requirements have accelerated the necessity of automation. This study presents a novel hybrid feature-selection technique for detecting network and device failure in IP core networks that connect 5 G core networks for internet connectivity. This approach first performs feature ranking using the MI (Mutual information), CC (Correlation coefficient), and GI (Gini index) to get the three different feature sets. Subsequently, in order to acquire a single optimized feature set, features are combined using a union operation. This optimized feature set is fed to the wrapper-based Boruta feature selection algorithm. In addition, to handle the data imbalance problem, we have applied the SMOTETomek method. Finally, obtained optimized feature set is fed to the three most popular ensemble approaches, Random forest, LGBM (Light gradient boosting machine), and xGBoost for the detection of failure. The proposed detection system's performance is evaluated using the KDDI dataset provided by KDDI Corporation. The effectiveness of the proposed FDF-HybridFS is tested and compared with several recent state-of-the-art methods cited in the related work in terms of detection rate, precision, accuracy, and F1 score. An analysis of the proposed failure detection framework's performance on the KDDI dataset shows that it outperforms state-of-the-art methods in terms of both detection rate and detection accuracy.

Reliable prediction of software defects using Shapley interpretable machine learning models

Predicting defect-prone software components can play a significant role in allocating relevant testing resources to fault-prone modules and hence increasing the business value of software projects. Most of the current software defect prediction studies utilize traditional supervised machine learning algorithms to predict defects in software applications. The software datasets utilized in such studies are imbalanced and therefore the reported results cannot be reliably used to judge their performance. Moreover, it is important to explain the output of machine learning models employed in fault-predication techniques to determine the contribution of each utilized feature to the model output. In this paper, we propose a new framework for predicting software defects utilizing eleven machine learning classifiers over twelve different datasets. For feature selection, we employ four different nature-inspired search algorithms, namely, particle swarm optimization, genetic algorithm, harmony algorithm, and ant colony optimization. Moreover, we make use of the synthetic minority oversampling technique (SMOTE) to address the problem of data imbalance. Furthermore, we utilize the Shapley additive explanation model for highlighting the highest determinative features. The obtained results demonstrate that gradient boosting, stochastic gradient boosting, decision trees, and categorical boosting outperform others tested model with over 90% accuracy and ROC-AUC. Additionally, we found that the ant colony optimization technique outperforms the other tested feature extraction techniques.

K-NEAREST NEIGHBOR DENGAN ADAPTIVE BOOSTING DAN SYNTHETIC MINORITY OVERSAMPLING TECHNIQUE UNTUK KLASIFIKASI DATA TIDAK SEIMBANG

Breast cancer is non-skin cancer that is caused by several factors, including glandular ducts, cells, and breast support tissue, except for the skin of the breast. Breast cancer if not treated immediately will be fatal for the sufferer, so early detection of breast cancer is important for the patient's safety. The success of breast cancer detection depends on the right diagnosis. Measurement of the accuracy of a breast cancer diagnosis can be assisted by statistical methods, namely classification. K-Nearest Neighbor is a classification algorithm based on the nearest neighbor that is easy to implement. In the classification process, there are several problems including when faced with imbalanced data. Imbalanced data can cause classification algorithms to tend to focus on the majority class. Data imbalance can be overcome by using Synthetic Minority Oversampling Technique (SMOTE). Ensemble methods can be applied to improve the performance of imbalanced data classification, one of which is Adaptive Boosting. This study applies K-Nearest Neighbor combined with Adaptive Boosting and SMOTE for handling imbalanced data classification. The results of this study are, SMOTE can handle the problem of imbalanced data and the application of K-Nearest Neighbor with Adaptive Boosting can produce an accuracy of 80%, a sensitivity of 83,33%, a specificity of 66,67%, and a G-Mean value of 74,54%. So it can be concluded that K-Nearest Neighbor combined with Adaptive Boosting and SMOTE can be applied for handling imbalanced data classification.

A novel multi-model cascade framework for pipeline defects detection based on machine vision

Defect detection technology is vital for ensuring the safety of pipelines during transportation. However, the current methods for defect detection using machine vision rely on having enough labeled defect samples. Unfortunately, some specific defect samples are difficult to obtain in engineering practice, which creates an imbalanced data problem and limits detection performance. Furthermore, traditional methods struggle to achieve satisfactory results with low-quality images. To solve these problems, a novel multi-model cascade framework based on machine vision is proposed. This framework uses a modified Super-Resolution Generative Adversarial Network (MSRGAN) with a self-attention mechanism to generate high-quality fake defect samples to balance data distribution. An improved Visual Geometry Group network (IVGG16) is also designed to enhance the performance of imbalanced defect classification, and Mask R-CNN is utilized to locate the defects. The experimental results demonstrate that the proposed framework performs well in recognizing imbalanced and low-quality samples, and it outperforms other state-of-the-art methods in terms of detection accuracy.

FundusGAN: Fundus image synthesis based on semi-supervised learning

Our goal is to construct a high-performance model that generates two types of fundus disease images for both Diabetic Retinopathy (DR) and Age-Related Macular degeneration (AMD) that have different symptoms for each severity for overcoming the lack of labeled data and data imbalance problems that hinder to build a model based on deep learning. To this end, we propose a framework named FundusGAN for building fundus image generators in semi-supervised learning. First, a semantic coarse lesion mask as a guidance mask is exploited to express disease and severity information more accurately to generate a final fundus image by adopting a coarse-to-fine strategy. Second, the disease-feature matching loss is proposed to learn the abundant features of the two diseases by employing the unlabeled datasets and is able to generate the initial symptoms of two fundus diseases accurately. We demonstrate the effectiveness of the proposed FundusGAN in comparison with carefully tuned baselines and state-of-the-art DR fundus image generation model and finally achieve 0.9174, and 0.8254 f1-score for DR and AMD, respectively.

RDDL: A systematic ensemble pipeline tool that streamlines balancing training schemes to reduce the effects of data imbalance in rare-disease-related deep-learning applications

Identifying lowly prevalent diseases, or rare diseases, in their early stages is key to disease treatment in the medical field. Deep learning techniques now provide promising tools for this purpose. Nevertheless, the low prevalence of rare diseases entangles the proper application of deep networks for disease identification due to the severe class-imbalance issue. In the past decades, some balancing methods have been studied to handle the data-imbalance issue. The bad news is that it is verified that none of these methods guarantees superior performance to others. This performance variation causes the need to formulate a systematic pipeline with a comprehensive software tool for enhancing deep-learning applications in rare disease identification. We reviewed the existing balancing schemes and summarized a systematic deep ensemble pipeline with a constructed tool called RDDL for handling the data imbalance issue. Through two real case studies, we showed that rare disease identification could be boosted with this systematic RDDL pipeline tool by lessening the data imbalance problem during model training. The RDDL pipeline tool is available at https://github.com/cobisLab/RDDL/.

Automatic microseismic signal classification for mining safety monitoring using the WaveNet classifier

AbstractMicroseismic monitoring is a promising method for the safety monitoring of underground mines. However, it is crucial to isolate microseismic signals related to the collapse of a mine from others for successful monitoring because the monitoring system records various signals. Recently, deep learning‐based classification techniques have achieved high performance in such data classification problems. In this context, we develop an automatic signal classification technique using the modified WaveNet classifier. The main characteristic of the WaveNet structure is its ability to extract features at various frequencies from very long time‐series data, and such an advantage makes the WaveNet suitable for seismic data processing. The data imbalance problem coming from the safe condition of the monitoring target is solved by augmenting the training data with those acquired from another mine and employing class weighting. After training, an optimal classifier is chosen considering the loss function, accuracy and Fβ score. The optimal classifier shows very high accuracy and excellent performance for the test data prediction. Compared to the random forest model and another one‐dimensional convolutional neural network–based network, the suggested classifier has higher reliability in predicting microseismic signals. Even though the proposed WaveNet model has a much more complex structure than the random forest model, the actual application examples demonstrate that the proposed model achieves high efficiency without any preprocessing. The automatic signal classifier developed in this study can be directly applied to various safety monitoring problems, not only mines, to improve the efficiency and reliability of monitoring systems.

A deep learning self-attention cross residual network with Info-WGANGP for mitotic cell identification in HEp-2 medical microscopic images

BackgroundThe identification of human epithelial type-2 mitotic cell patterns in the indirect immunofluorescence images (IIF HEp-2) is a critical step for autoimmune diseases computer-aided diagnosis (CAD) systems. Recognition of HEp-2 cells in the mitotic phase is clinically vital for validating the HEp-2 samples and assisting in diagnosing specimens with mixed patterns. Typically, mitotic cells are rarely observed in the HEp-2 specimen images, resulting in a significant skewness of the available medical datasets towards the majority of interphase (non-mitotic) patterns. MethodsIn this paper, a deep learning framework is proposed based on a self-attention deep cross-residual network with an efficient generative adversarial network (GAN). The proposed framework consists of two consecutive steps. First, the problem of imbalanced data of minority mitotic against the majority interphase cells is remedied using Info-WGANGP approach in combination with the conventional data augmentation methods. Second, a downstream end-to-end deep learning Att-DCRNet is developed to classify the mitotic and interphase cell patterns of the IIF HEp-2 medical images. A comprehensive experimental study is performed to validate the effectiveness of the proposed framework against other state-of-the-art methods over the public medical dataset of UQ-SNP_HEp-2 Task-3. ResultsThe proposed framework demonstrates competitive classification results attaining a maximum performance of 84.10% F1-score, 84.70% Matthew's correlation coefficient (MCC), and 99.0% balanced class accuracy (BcA), which proves its applicability for automatically supporting an accurate diagnosis decision regarding the HEp-2 mitotic and interphase cell patterns. The source code is available at this link: https://github.com/Anaam-dl/AttDCRNet_with_InfoWGANGP.

CMTSNN: A Deep Learning Model for Multiclassification of Abnormal and Encrypted Traffic of Internet of Things

With the increasing types and number of IoT devices and malicious programs and the popularization of encryption technology in the communication process between the internet and the Internet of Things (IoT), a large amount of encrypted abnormal traffic among devices endangers IoT cybersecurity. How to identify abnormal encrypted traffic of the IoT has become the premise of cybersecurity. Presently, most of the detection methods for traffic in the IoT have problems such as simple dataset processing, imperfect feature extraction, data imbalance, and low multi-classification accuracy. In this paper, we propose a multi-classification deep learning model named the cost matrix time-space neural network (CMTSNN) for abnormal and encrypted IoT traffic. The CMTSNN is divided into three parts. The first part is the preprocessing stage of the dataset, which needs to retain the timing relation between two data packets in the stream and create a cost penalty matrix according to the sample distribution. Aimed at the robustness of feature extraction in network flow, the second part extracts time series features and then space features to ensure the robustness of feature extraction. The third part is aimed at the problem of data imbalance. The cost penalty matrix is applied to the cost penalty layer in the training process, and then the improved cross-entropy loss function is used to calculate the loss to improve the classification accuracy of minority categories and increase the overall multi-classification performance of the model. Experiments were carried out with the ToN-IoT, BoT-IoT and ISCX VPN-NonVPN datasets. Compared with current methods, the proposed method shows better performances, including accuracy, precision, recall, F1 Score and false alarm rate.

Semi-supervised hybrid spine network for segmentation of spine MR images.

Automatic segmentation of vertebral bodies (VBs) and intervertebral discs (IVDs) in 3D magnetic resonance (MR) images is vital in diagnosing and treating spinal diseases. However, segmenting the VBs and IVDs simultaneously is not trivial. Moreover, problems exist, including blurry segmentation caused by anisotropy resolution, high computational cost, inter-class similarity and intra-class variability, and data imbalances. We proposed a two-stage algorithm, named semi-supervised hybrid spine network (SSHSNet), to address these problems by achieving accurate simultaneous VB and IVD segmentation. In the first stage, we constructed a 2D semi-supervised DeepLabv3+ by using cross pseudo supervision to obtain intra-slice features and coarse segmentation. In the second stage, a 3D full-resolution patch-based DeepLabv3+ was built. This model can be used to extract inter-slice information and combine the coarse segmentation and intra-slice features provided from the first stage. Moreover, a cross tri-attention module was applied to compensate for the loss of inter-slice and intra-slice information separately generated from 2D and 3D networks, thereby improving feature representation ability and achieving satisfactory segmentation results. The proposed SSHSNet was validated on a publicly available spine MR image dataset, and remarkable segmentation performance was achieved. Moreover, results show that the proposed method has great potential in dealing with the data imbalance problem. Based on previous reports, few studies have incorporated a semi-supervised learning strategy with a cross attention mechanism for spine segmentation. Therefore, the proposed method may provide a useful tool for spine segmentation and aid clinically in spinal disease diagnoses and treatments. Codes are publicly available at: https://github.com/Meiyan88/SSHSNet.

Sentiment Analysis of Shopee Food Application User Satisfaction Using the C4.5 Decision Tree Method

Sentiment analysis on public opinion regarding the shopee food application is an interesting topic in the context of evaluating service quality in the shopee food application. In this digital era, user opinion has a very important role in shaping public perception of the application. Therefore, sentiment analysis is needed to understand user opinion about the shopee food application. This study uses Decision Tree C4.5 to analyze public sentiment on the use of the Shopee Food application on Twitter users. However, beforehand it is necessary to overcome the problem of data imbalance which is common in datasets, where the number of positive, negative, and neutral sentiments is not balanced. To overcome this problem, three different techniques are used, namely SMOTE, undersampling, and a combination of oversampling and undersampling. The results of this study indicate that the SMOTE technique provides better results in overcoming data imbalances and increasing prediction accuracy. With an accuracy of 0.88. the SMOTE technique can provide more accurate sentiment predictions than the undersampling technique and the combination of oversampling and undersampling. This is because SMOTE can synthetically expand the number of minority samples, thereby preventing the loss of information and maintaining variation in the dataset. In conclusion, sentiment analysis on the Shopee Food application on Google Play using the Decision Tree C4.5 algorithm and the SMOTE technique can overcome data imbalances with a prediction accuracy of 0.88. This technique is more efficient than the undersampling technique and the combination of oversampling and undersampling. These results can provide developers with valuable insights to improve app quality and user satisfaction.

A Study on Heuristic Algorithms Combined With LR on a DNN-Based IDS Model to Detect IoT Attacks

Current security challenges are made more difficult by the complexity and difficulty of spotting cyberattacks due to the Internet of Things explosive growth in connected devices and apps. Therefore, various sophisticated attack detection techniques have been created to address these issues in recent years. Due to their effectiveness and scalability, machine learning-based intrusion detection systems (IDSs) have increased. However, several factors, such as the characteristics of the training dataset and the training model, affect how well these AI-based systems identify attacks. In particular, the heuristic algorithms (GA, PSO, CSO, FA) optimized by the logistic regression (LR) approach employ it to pick critical features of a dataset and deal with data imbalance problems. This study offers an intrusion detection system (IDS) based on a deep neural network and heuristic algorithms combined with LR to boost the accuracy of attack detections. Our proposed model has a high attack detection rate of up to 99% when testing on the IoT-23 dataset.

A Review on Oversampling Techniques for Solving the Data Imbalance Problem in Classification

The data imbalance problem is a widely explored area in the Machine Learning domain. With the rapid advancement of computing infrastructure and the incessant increase in the amount and variety of data generated, the data imbalance problem has prevailed and reshaped with the requirement for novel approaches to address it. Among the different approaches that exist to address the data imbalance problem, such as data-level and algorithmic-level, data-level approaches are more popular among the scientific community due to their classifier-independent nature. When investigating current trends in data-level approaches, it is evident that oversampling is a technique frequently explored due to its adaptability to scenarios where extreme data imbalance is present. This paper presents a review of different oversampling techniques with a comprehensive analysis of the strategies that have been used along with possible areas that looks promising to explore further to develop more advanced oversampling techniques.

Real-World Deep Local Motion Deblurring

Most existing deblurring methods focus on removing global blur caused by camera shake, while they cannot well handle local blur caused by object movements. To fill the vacancy of local deblurring in real scenes, we establish the first real local motion blur dataset (ReLoBlur), which is captured by a synchronized beam-splitting photographing system and corrected by a post-progressing pipeline. Based on ReLoBlur, we propose a Local Blur-Aware Gated network (LBAG) and several local blur-aware techniques to bridge the gap between global and local deblurring: 1) a blur detection approach based on background subtraction to localize blurred regions; 2) a gate mechanism to guide our network to focus on blurred regions; and 3) a blur-aware patch cropping strategy to address data imbalance problem. Extensive experiments prove the reliability of ReLoBlur dataset, and demonstrate that LBAG achieves better performance than state-of-the-art global deblurring methods and our proposed local blur-aware techniques are effective.

LeafNet: Using Convolutional Neural Network for Plant Leaf Detection

Pakistan is home to thousands of plant species. As a result of pollution, natural disasters, and climate change, many of these species are at risk of extinction. Plant categorization and detection systems are designed to assist non-experts in automatically identifying plants based on their leaves to ensure their safety. The current study proposes a plant leaf detection system utilizing a Convolutional Neural Network architecture. Making use of the PlantVillage dataset, the proposed system can identify seven species of plants namely apple, cherry, tomato, potato, soybean, strawberry, and corn. Data augmentation strategies have been used to provide more training examples to get around the problem of bias and imbalanced data. The accuracy achieved on the training set was 98.87% which improved to 99.5% when using data augmentation. Apart from the monitoring of endangered species, the adoption of the proposed model can also aid the evaluation of weed management efforts and analysis of species distribution under climate change.

Self-Training with Entropy-Based Mixup for Low-Resource Chest X-ray Classification

Deep learning-based medical image analysis technology has been developed to the extent that it shows an accuracy surpassing the ability of a human radiologist in some tasks. However, data labeling on medical images requires human experts and a great deal of time and expense. Moreover, medical image data usually have an imbalanced distribution for each disease. In particular, in multilabel classification, learning with a small number of labeled data causes overfitting problems. The model easily overfits the limited number of labeled data, while it still underfits the large amount of unlabeled data. In this study, we propose a method that combines entropy-based Mixup and self-training to improve the performance of data-imbalanced chest X-ray classification. The proposed method is to apply the Mixup algorithm to limited labeled data to alleviate the data imbalance problem and perform self-training that effectively utilizes the unlabeled data while iterating this process by replacing the teacher model with the student model. Experimental results in an environment with a limited number of labeled data and a large number of unlabeled data showed that the classification performance was improved by combining entropy-based Mixup and self-training.

Anomaly Detection of Breast Cancer Using Deep Learning.

Cancer is one of the deadliest diseases facing humanity, one of the which is breast cancer, and it can be considered one of the primary causes of death for most women. Early detection and treatment can significantly improve outcomes and reduce the death rate and treatment costs. This article proposes an efficient and accurate deep learning-based anomaly detection framework. The framework aims to recognize breast abnormalities (benign and malignant) by considering normal data. Also, we address the problem of imbalanced data, which can be claimed to be a popular issue in the medical field. The framework consists of two stages: (1) data pre-processing (i.e., image pre-processing); and (2) feature extraction through the adoption of a MobileNetV2 pre-trained model. After that classification step, a single-layer perceptron is used. Two public datasets were used for the evaluation: INbreast and MIAS. The experimental results showed that the proposed framework is efficient and accurate in detecting anomalies (e.g., 81.40% to 97.36% in terms of area under the curve). As per the evaluation results, the proposed framework outperforms recent and relevant works and overcomes their limitations.

Respiratory Sound Classification by Applying Deep Neural Network with a Blocking Variable

Respiratory diseases are leading causes of death worldwide, and failure to detect diseases at an early stage can threaten people’s lives. Previous research has pointed out that deep learning and machine learning are valid alternative strategies to detect respiratory diseases without the presence of a doctor. Thus, it is worthwhile to develop an automatic respiratory disease detection system. This paper proposes a deep neural network with a blocking variable, namely Blnet, to classify respiratory sound, which integrates the strength of the ResNet, GoogleNet, and the self-attention mechanism. To solve the non-IID data problem, a two-stage training process with the blocking variable was developed. In addition, the mix-up data augmentation within the clusters was used to address the imbalanced data problem. The performance of the Blnet was tested on the ICBHI 2017 data, and the model achieved 79.13% specificity and 66.31% sensitivity, with an average score of 72.72%, which is a 4.22% improvement in the average score and a 12.61% improvement in sensitivity over the state-of-the-art results.

Recognition and location of marine animal sounds using two-stream ConvNet with attention

There are abundant resources and many endangered marine animals in the ocean. Using sound to effectively identify and locate them, and estimate their distribution area, has a very important role in the study of the complex diversity of marine animals (Hanny et al., 2013). We design a Two-Stream ConvNet with Attention (TSCA) model, which is a two-stream model combined with attention, in which one branch processes the temporal signal and the other branch processes the frequency domain signal; It makes good use of the characteristics of high time resolution of time domain signal and high recognition rate of frequency domain signal features of sound, and it realizes rapid localization and recognition of sound of marine species. The basic network architecture of the model is YOLO (You Only Look Once) (Joseph et al., 2016). A new loss function focal loss is constructed to strengthen the impact on the tail class of the sample, overcome the problem of data imbalance and avoid over fitting. At the same time, the attention module is constructed to focus on more detailed sound features, so as to improve the noise resistance of the model and achieve high-precision marine species identification and location. In The Watkins Marine Mammal Sound Database, the recognition rate of the algorithm reached 92.04% and the positioning accuracy reached 78.4%.The experimental results show that the algorithm has good robustness, high recognition accuracy and positioning accuracy.

Mineração de dados no ranqueamento de lotes de sementes de sorgo

ABSTRACT The ranking of seed lots is a fundamental process for all companies in the seed industry. This work aims to demonstrate data mining methods for ranking sorghum seed lots during the seed processing through analysis of quality control data. Germination and cold tests were performed to verify the physiological quality of the lots. Seed samples from each lot were evaluated in two moments: post-cleaning and finished product (ready for marketing). The results after pre-processing totaled 188 rows of data with six attributes, encompassing 150 lots accepted for marketing, 6 rejected, and 32 intermediate lots. The classifiers used were J48, Random Forest, Classification Via Regression, Naive Bayes, Multilayer Perceptron, and IBk. The Resample filter was used for adjustment of the data. The k-fold technique was used for training, with ten folds. The metrics of Accuracy, Precision, Recall, F-measure, and ROC Area were used to verify the accuracy of the algorithms. The results obtained were used to determine the best machine-learning algorithm. IBk and J48 presented the highest accuracy of data; the IBk technique presented the best results. The Resample filter was essential for solving the data imbalance problem. Sorghum seed lots can be classified with great accuracy and precision through artificial intelligence and machine learning technique.