Article
Computer Science, Artificial Intelligence
Bhagat Singh Raghuwanshi, Sanyam Shukla
Summary: This paper introduces a new variant of extreme learning machine, MCVCSELM, for effectively addressing binary class imbalance problems by utilizing minimum class variance and class-specific regularization. Experimental results demonstrate that the proposed algorithm outperforms several state-of-the-art methods for imbalanced learning.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Review
Engineering, Industrial
Andrea de Giorgio, Gabriele Cola, Lihui Wang
Summary: This article provides a systematic review of data manipulation, machine learning, and deep learning solutions to the class imbalance problem in the manufacturing domain. It critically evaluates different metrics and explores the availability of public source code and imbalanced datasets for benchmarking. Furthermore, it summarizes the most applied solutions to the class imbalance problem in manufacturing and discusses future challenges.
JOURNAL OF MANUFACTURING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Yanjiao Li, Jie Zhang, Sen Zhang, Wendong Xiao, Zhiqiang Zhang
Summary: This paper presents a multi-objective optimization-based adaptive class-specific cost extreme learning machine (MOAC-ELM) method for imbalanced classification problems. By considering the costs of different classes and enhancing the representation of the minority class using penalty factors, the class-specific costs are automatically determined. The proposed MOAC-ELM shows good robustness and generalization performance in imbalanced classification tasks, as demonstrated by comprehensive experiments.
Article
Computer Science, Information Systems
Shanlin Zhou, Yan Gu, Hualong Yu, Xibei Yang, Shang Gao
Summary: This article introduces a novel strategy called RUE for estimating location information and cost assignment to address the problem of class-imbalance learning. The strategy indirectly explores location information through a random undersampling ensemble, is robust towards data distribution, and accurately estimates the significance of each instance.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Bhagat Singh Raghuwanshi
Summary: The conventional extreme learning machine (ELM) fails to handle the class imbalance problem effectively because it treats all samples as equally important. To address this issue, modified versions of ELM like weighted ELM (WELM) and overall distribution WELM (ODW-ELM) have been developed. In this study, a class-specific ELM based on overall distribution (OD-CSELM) and its kernelized version (OD-CSKELM) are proposed to handle binary class imbalance problem more effectively. OD-CSELM and OD-CSKELM have lower computational complexity compared to WELM and kernelized WELM, respectively. Experimental results on benchmark datasets demonstrate the superior generalization performance of the proposed methods for class imbalance learning.
Article
Computer Science, Artificial Intelligence
Roshani Choudhary, Sanyam Shukla
Summary: This paper proposes an ensemble method that decomposes a complex imbalanced problem into simpler sub-problems, solves them using cost-sensitive classifiers, and combines the results using voting methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Qi Dai, Jian-wei Liu, Yong-hui Shi
Summary: The class-imbalance problem is a significant issue in machine learning and data mining. Many methods have been developed to address this problem, but they usually overlook global similarity. To find the global similarity of datasets, a novel Schur decomposition class-overlap undersampling method (SDCU) is proposed. Experimental results demonstrate the superior performance of SDCU compared to other state-of-the-art methods on various classifiers.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Leonardo Canete-Sifuentes, Raul Monroy, Miguel Angel Medina-Perez
Summary: Decision trees are popular classifiers known for their good performance, interpretability, and use in ensembles. This paper introduces a new functional tree, FT4cip, designed to address the issue of class imbalance in databases. The experimental results demonstrate that FT4cip outperforms the best model tree and functional tree in terms of AUC. The study also includes meta-analysis and comparison with other models to recommend classifiers based on different database types.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Computer Science, Information Systems
Jinjun Ren, Yuping Wang, Xiyan Deng
Summary: Class imbalance and noisy data pose challenges for constructing good classifiers using SVM. Fuzzy SVMs (FSVMs) address these issues by using fuzzy membership functions and cost-sensitive learning. However, the accuracy of FSVMs is affected by class imbalance. To overcome this, we propose SFFSVM, which incorporates a new fuzzy membership function and adjusts the importance of samples based on the relationship between estimated and optimal hyperplanes. Experimental results show that SFFSVM outperforms other methods on F1, MCC, and AUC-PR metrics.
ACM TRANSACTIONS ON KNOWLEDGE DISCOVERY FROM DATA
(2023)
Article
Computer Science, Artificial Intelligence
M. A. Ganaie, M. Tanveer, Alzheimer's Disease Neuroimaging Initiative
Summary: In real world problems, the imbalance of data samples presents a challenge for classification models. This paper proposes a K-nearest neighbor based weighted reduced universum twin support vector machine model to address class imbalance issues by incorporating local neighborhood information and utilizing universum data, resulting in improved generalization performance.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Automation & Control Systems
Qi Dai, Jian-wei Liu, Jia-Peng Yang
Summary: For class-imbalance problems, traditional supervised learning algorithms struggle with accurately identifying minority instances. Ensemble learning, by building multiple classifiers on the training dataset, can improve recognition accuracy for minority instances. Few researchers have used sliding windows to select majority instances and construct ensemble learning models. Therefore, this paper proposes a novel sliding window-based selective ensemble learning method (SWSEL) that utilizes similarity mapping and distance alignment to address the class-imbalance problem.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Review
Computer Science, Information Systems
Ahmad S. Tarawneh, Ahmad B. Hassanat, Ghada Awad Altarawneh, Abdullah Almuhaimeed
Summary: Oversampling has been used to address the challenge of learning from imbalanced datasets, but it also has its downsides. The synthesized samples may not truly represent the minority class, resulting in incorrect predictions in real-world applications. This paper analyzes various oversampling methods and proposes a new evaluation system based on comparing hidden majority examples with those generated by oversampling. Experimental results show that all studied oversampling methods produce minority samples that are most likely to be majority. Given the data and methods at hand, oversampling in its current forms and methodologies is deemed unreliable and should be avoided in real-world applications.
Article
Computer Science, Artificial Intelligence
Paria Soltanzadeh, M. Reza Feizi-Derakhshi, Mahdi Hashemzadeh
Summary: This research proposes an under-sampling approach based on a metaheuristic method to address the problems of imbalanced class distribution and class overlap, achieving significant performance improvement compared to competitors.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Behzad Mirzaei, Farshad Rahmati, Hossein Nezamabadi-pour
Summary: This paper proposes a score-based preprocessing technique based on under-sampling and over-sampling to overcome the weakness of classifiers in class imbalance problems. The technique selects suitable samples based on their importance in the feature space and balances the classes' distribution. Experiments show the superiority and effectiveness of the proposed method compared to other methods.
PATTERN ANALYSIS AND APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Roozbeh Razavi-Far, Maryam Farajzadeh-Zanajni, Boyu Wang, Mehrdad Saif, Shiladitya Chakrabarti
Summary: The correct classification of rare samples is crucial and this article proposes novel oversampling strategies based on imputation methods to address this issue. The techniques are designed to generate synthetic minority class samples and outperform other methods according to performance metrics such as AUC, F-measure, and G-mean.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2021)