Class imbalances versus small disjuncts

By Taeho C. Jo, Nathalie Japkowicz. SKDD, 2004

747 citations recorded by the source corpus.

Subfields: Accounting, Artificial Intelligence, Electrical and Electronic Engineering

View Paper: https://www.semanticscholar.org/paper/09ddfea0d38687d88780a72a583095274fd0510f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Accounting

In Artificial Intelligence

In Electrical and Electronic Engineering

Citations in the corpus

Addressing Class Imbalance in Federated Learning (2020, methodology): “In the literature, a number of approaches have been proposed to address class imbalance, e.g., applying various data sampling techniques (Jo and Japkowicz 2004), using generative augmentation to make up for the lack of minority samples (Lee, Park, and Kim 2016; Pouyanfar et al. 2018), and…”
A Survey of Predictive Modeling on Imbalanced Domains (2016, methodology): “The relation between imbalanced domains and small sample problems was addressed by Japkowicz and Stephen [2002] and Jo and Japkowicz [2004], where it was highlighted that minority class examples are easier to learn as their number increases.”
Deep learning in bioinformatics (2016, methodology): “Sampling methods balance the distribution of imbalanced data, and several approaches have been proposed, including informed undersampling [184], the synthetic minority oversampling technique [185] and cluster-based sampling [186].”
FSVM-CIL: Fuzzy Support Vector Machines for Class Imbalance Learning (2010, methodology): “A clustering-based sampling method has been proposed in [22], while a genetic algorithm-based sampling method has been proposed in [10].”
Training and assessing classification rules with imbalanced data (2012, background): “Anargumentthatmovesawayfromtheonesjustmentioned,suggeststhattheeffects of class imbalance are not directly caused by the distribution of the classes, but rather that the imbalance may lead to small disjuncts determining, in turn, a degradation of the classiﬁcation (Jo and Japkowicz 2004).”
Survey of resampling techniques for improving classification performance in unbalanced datasets (2016, methodology): “Some of these include One side selection [9], ADASYN [5], SVM SMOTE [12], SMOTEBoost [2], ClusterBased Oversampling [7], Kernel-based methods and active learning [6].”
An insight into classification with imbalanced data: Empirical results and current trends on using data intrinsic characteristics (2013, methodology): “Again , the application of the CBO approach results on a per - fect classification for all data , having 7 .”
Cluster-based under-sampling approaches for imbalanced data distributions (2009, background): “However, many datasets in real applications involve imbalanced class distribution problem (Chawla, 2003; Chyi, 2003; Japkowicz, 2000, 2001; Jo & Japkowicz, 2004; Maloof, 2003; Zhang & Mani, 2003).”
A novel ensemble method for classifying imbalanced data (2015, background): “Apart from the class imbalance characteristic, there might be other data characteristics that could affect the performance of traditional classification algorithms for imbalanced problems [45–47], such as dataset shift [48], class overlapping [49], and small disjuncts [50].”
Data Mining for Imbalanced Datasets: An Overview (2005, background): “…significant attention to counter the effect of imbalanced data sets (Solberg and Solberg, 1996; Japkowicz , 2000a; Chawla et al., 2002; Weiss and Provost, 2003; Kubat and Matwin, 1997; Jo and Japkowicz, 2004; Batista et al., 2004; Phua and Alahakoon , 2004; Laurikkala, 2001; Ling and Li, 1998).”
GraphSMOTE: Imbalanced Node Classification on Graphs with Graph Neural Networks (2021, background): “Over-sampling scale is varied as { 0 .”
Oversampling for Imbalanced Learning Based on K-Means and SMOTE (2017, methodology): “A simple and common approach to this problem is to use stratiﬁed cross-validation, where instead of sampling completely at random, the original class distribution is preserved in each fold (Japkowicz, 2013).”
SMOTE-IPF: Addressing the noisy and borderline examples problem in imbalanced classification by a re-sampling method with filtering (2015, methodology): “Presence of small disjuncts [28,29] (Fig.”
A systematic study of the class imbalance problem in convolutional neural networks (2017, background): “Cluster-based oversampling ﬁrst clusters the dataset and then oversamples each cluster separately [36].”
Borderline-SMOTE: A New Over-Sampling Method in Imbalanced Data Sets Learning (2005, background): “with between-class imbalance and within-class imbalance simultaneously [16].”
Borderline over-sampling for imbalanced data classification (2009, methodology): “Jo and Japkowicz [11] partitioned the minority and majority classes separately using the k-means algorithm, then oversampled the clusters by duplication in order to re-balance class distribution as well as inflating small disjuncts of the minority class to counter within-class imbalance.”
Interpreting and Unifying Outlier Scores (2011, methodology): “In the supervised context of classification, much effort has been spent on methods of sanitizing imbalanced class prior probabilities for training as well as for evaluation of classifiers [7,24,44,53].”
Classification with class imbalance problem: A review (2015, background): “The studies of [35] and [50] explored the within class imbalance in minority class and claimed that the underrepresented minority class caused by small disjunct could be improved by applying a guided upsampling in respect to the minority class.”