To Combat Multi-Class Imbalanced Problems by Means of Over-Sampling Techniques

Abdi, Lida; Hashemi, Sattar

doi:10.1109/tkde.2015.2458858

Cited by 321 publications

(121 citation statements)

References 37 publications

(35 reference statements)

Supporting

Mentioning

108

Contrasting

Unclassified

Order By: Relevance

“…-New sampling strategies are required for multi-class problems. Simple re-balancing towards the biggest or smallest class is not a proper approach [1,15]. We need to develop dedicated methods that will adjust the sampling procedures to both individual properties of classes and to their mutual relations.…”

Section: Data Preprocessingmentioning

confidence: 99%

Learning from imbalanced data: open challenges and future directions

2016

View full text Add to dashboard Cite

Despite more than two decades of continuous development learning from imbalanced data is still a focus of intense research. Starting as a problem of skewed distributions of binary tasks, this topic evolved way beyond this conception. With the expansion of machine learning and data mining, combined with the arrival of big data era, we have gained a deeper insight into the nature of imbalanced learning, while at the same time facing new emerging challenges. Data-level and algorithm-level methods are constantly being improved and hybrid approaches gain increasing popularity. Recent trends focus on analyzing not only the disproportion between classes, but also other difficulties embedded in the nature of data. New real-life problems motivate researchers to focus on computationally efficient, adaptive and real-time methods. This paper aims at discussing open issues and challenges that need to be addressed to further develop the field of imbalanced learning. Seven vital areas of research in this topic are identified, covering the full spectrum of learning from imbalanced data: classification, regression, clustering, data streams, big data analytics and applications, e.g., in social media and computer vision. This paper provides a discussion and suggestions concerning lines of future research for each of them.

show abstract

Section: Data Preprocessingmentioning

confidence: 99%

Learning from imbalanced data: open challenges and future directions

2016

View full text Add to dashboard Cite

show abstract

“…Table 2 presents a confusion matrix [28] for a binary class that indicates correct and incorrect classifications. Previous studies [29,30] show that accuracy is not an appropriate measure to evaluate the performance of classifying imbalanced dataset. This is because it only considers a number of correctly classified instances but does not pay attention to how many of them are of a positive class which is important for us.…”

Section: Evaluation Parametersmentioning

confidence: 99%

A Hybrid Approach for Class Imbalance Problem in Customer Churn Prediction: A Novel Extension to Under-sampling

Salunkhe¹,

Mali²

2018

IJISA

View full text Add to dashboard Cite

Abstract-Customer retention is becoming a key success factor for many business applications due to increasing market competition. Especially telecom companies are facing this challenge with a rapidly increasing number of service providers. Hence there is need to focus on customer churn prediction in order to detect the customers that are likely to churn i.e. switch from one service provider to another. Several data mining techniques are applied for classifying customers into the churn and non-churn category. But churn prediction applications comprise an imbalanced distribution of the dataset.One of the commonly used techniques to handle imbalanced data is re-sampling of data as it is independent of the classifier being used. In this paper, we develop a hybrid re-sampling approach named SOS-BUS by combining well known oversampling technique SMOTE with our novel under-sampling technique. Our methodology aims to focus on the necessary data of majority class and avoid their removal in order to overcome the limitation of random under-sampling. Experimental results show that the proposed approach outperforms the other reference techniques in terms of Area under ROC Curve (AUC).

show abstract

“…However, many of the resampling and data augmentation techniques that have been proven to be effective for binary classification problems (such as SMOTE [11] and ADASYN [28]) can actually impair classifier performance when applied to multiclass problems [3]. Some research into resampling techniques for multi-class problems does exist [20], [3], but the majority of experimental results only consider low dimensional data sets.…”

Section: B Class Imbalancementioning

confidence: 99%

“…As mentioned in Section II-B imbalanced data sets, and particularly those with multiple classes, introduce many difficulties in the training of machine learning algorithms [50]. This is compounded by the fact that many of the data augmentation techniques that are effective in binary classification problems can do more harm than good when applied to multi-class problems [3].…”

Section: A Problem Descriptionmentioning

confidence: 99%

Multi-objective evolution of artificial neural networks in multi-class medical diagnosis problems with class imbalance

Shenfield

Rostami

2017

2017 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB)

View full text Add to dashboard Cite

Abstract-This paper proposes a novel multi-objective optimisation approach to solving both the problem of finding good structural and parametric choices in an ANN and the problem of training a classifier with a heavily skewed data set. The state-of-the-art CMA-PAES-HAGA multi-objective evolutionary algorithm [41] is used to simultaneously optimise the structure, weights, and biases of a population of ANNs with respect to not only the overall classification accuracy, but the classification accuracies of each individual target class. The effectiveness of this approach is then demonstrated on a real-world multi-class problem in medical diagnosis (classification of fetal cardiotocogorams) where more than 75% of the data belongs to the majority class and the rest to two other minority classes. The optimised ANN is shown to significantly outperform a standard feed-forward ANN with respect to minority class recognition at the cost of slightly worse performance in terms of overall classification accuracy.

show abstract

To Combat Multi-Class Imbalanced Problems by Means of Over-Sampling Techniques

Cited by 321 publications

References 37 publications

Learning from imbalanced data: open challenges and future directions

Learning from imbalanced data: open challenges and future directions

A Hybrid Approach for Class Imbalance Problem in Customer Churn Prediction: A Novel Extension to Under-sampling

Multi-objective evolution of artificial neural networks in multi-class medical diagnosis problems with class imbalance

Contact Info

Product

Resources

About