Classification for Imbalanced and Overlapping Classes Using Outlier Detection and Sampling Techniques

Zeping, Yang; Gao, Daqi

doi:10.12785/amis/071l50

Cited by 13 publications

(16 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the contrary, adding synthetic instances to the minority class (SMOTE) may introduce noise along the borderlines, leading to the loss of activity cliffs and mislabeling of the synthetic instances [ 66 ]. The ENN algorithm may effectively remove those synthetic outliers and restore the activity cliffs and class boundaries, leading to enhanced prediction performance for SMOTEENN (SMN) [ 67 ]…”

Section: Resultsmentioning

confidence: 99%

Structure–activity relationship-based chemical classification of highly imbalanced Tox21 datasets

et al. 2020

View full text Add to dashboard Cite

The specificity of toxicant-target biomolecule interactions lends to the very imbalanced nature of many toxicity datasets, causing poor performance in Structure–Activity Relationship (SAR)-based chemical classification. Undersampling and oversampling are representative techniques for handling such an imbalance challenge. However, removing inactive chemical compound instances from the majority class using an undersampling technique can result in information loss, whereas increasing active toxicant instances in the minority class by interpolation tends to introduce artificial minority instances that often cross into the majority class space, giving rise to class overlapping and a higher false prediction rate. In this study, in order to improve the prediction accuracy of imbalanced learning, we employed SMOTEENN, a combination of Synthetic Minority Over-sampling Technique (SMOTE) and Edited Nearest Neighbor (ENN) algorithms, to oversample the minority class by creating synthetic samples, followed by cleaning the mislabeled instances. We chose the highly imbalanced Tox21 dataset, which consisted of 12 in vitro bioassays for > 10,000 chemicals that were distributed unevenly between binary classes. With Random Forest (RF) as the base classifier and bagging as the ensemble strategy, we applied four hybrid learning methods, i.e., RF without imbalance handling (RF), RF with Random Undersampling (RUS), RF with SMOTE (SMO), and RF with SMOTEENN (SMN). The performance of the four learning methods was compared using nine evaluation metrics, among which F1 score, Matthews correlation coefficient and Brier score provided a more consistent assessment of the overall performance across the 12 datasets. The Friedman’s aligned ranks test and the subsequent Bergmann-Hommel post hoc test showed that SMN significantly outperformed the other three methods. We also found that a strong negative correlation existed between the prediction accuracy and the imbalance ratio (IR), which is defined as the number of inactive compounds divided by the number of active compounds. SMN became less effective when IR exceeded a certain threshold (e.g., > 28). The ability to separate the few active compounds from the vast amounts of inactive ones is of great importance in computational toxicology. This work demonstrates that the performance of SAR-based, imbalanced chemical toxicity classification can be significantly improved through the use of data rebalancing.

show abstract

Section: Resultsmentioning

confidence: 99%

Structure–activity relationship-based chemical classification of highly imbalanced Tox21 datasets

et al. 2020

View full text Add to dashboard Cite

show abstract

“…In addition to the PRISM algorithm [51], other outlier and/or noise detection methods have been recently introduced demonstrating their ability to improve the prediction accuracy of machine learning models [53,54]. For example, Byeon et al introduced a novel technique to enhance the quality of training data with a noisy dependent variable for binary classification [53].…”

Section: Activity-cliff-related Medicinal Chemistry Task Approach Refsmentioning

confidence: 99%

“…More recently, based on the borderline noise factor, Yang and Gao applied datacleaning techniques to remove the classification borderline noise. This work compared three under-sampling methods to select the representative majority class examples and remove the distant samples, which are useless to form the decision boundary [54]. The experimental results on bench datasets showed that the proposed method can effectively improve the classification accuracy of minority classes while achieving better overall classification.…”

Section: Activity-cliff-related Medicinal Chemistry Task Approach Refsmentioning

confidence: 99%

“…compounds) is a well-justified practice in machine learning to improve the prediction accuracy of models [51,53,54,56], it follows that it is reasonable to remove activity cliffs (smoothing out the activity landscape) for developing predictive QSAR models. The obvious drawback of this procedure is the inevitable loss of potentially crucial SAR information.…”

Section: Activity-cliff-related Medicinal Chemistry Task Approach Refsmentioning

confidence: 99%

See 1 more Smart Citation

Activity cliffs in drug discovery: Dr Jekyll or Mr Hyde?

Cruz-Monteagudo¹,

Medina-Franco²,

Pérez-Castillo³

et al. 2014

Drug Discovery Today

153

134

View full text Add to dashboard Cite

“…Another reason of overlapping is because the samples in both classes share almost that same value of attributes. Such overlapping causes difficulties for a classifier to classify the samples and may eventually lead to a low classification rate [49], [50]. In the next section, solutions to the unbalanced class distribution and the overlapping class samples shall be discussed based on the previous work conducted by other researchers.…”

Section: Overlapping Of Classesmentioning

confidence: 99%

A Multiple Classifiers System for Anomaly Detection in Credit Card Data With Unbalanced and Overlapped Classes

Kalid

Tong

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Frauds and default payments are two major anomalies in credit card transactions. Researchers have been vigorously finding solutions to tackle them and one of the solutions is to use data mining approaches. However, the collected credit card data can be quite a challenge for researchers. This is because of the data characteristics that contain: (i) unbalanced class distribution, and (ii) overlapping of class samples. Both characteristics generally cause low detection rates for the anomalies that are minorities in the data. On top of that, the weakness of general learning algorithms contributes to the difficulties of classifying the anomalies as the algorithms generally bias towards the majority class samples. In this study, we used a Multiple Classifiers System (MCS) on these two data sets: (i) credit card frauds (CCF), and (ii) credit card default payments (CCDP). The MCS employs a sequential decision combination strategy to produce accurate anomaly detection. Our empirical studies show that the MCS outperforms the existing research, particularly in detecting the anomalies that are minorities in these two credit card data sets. INDEX TERMS Anomaly detection, credit card, multiple classifiers.

show abstract

Classification for Imbalanced and Overlapping Classes Using Outlier Detection and Sampling Techniques

Cited by 13 publications

References 1 publication

Structure–activity relationship-based chemical classification of highly imbalanced Tox21 datasets

Structure–activity relationship-based chemical classification of highly imbalanced Tox21 datasets

Activity cliffs in drug discovery: Dr Jekyll or Mr Hyde?

A Multiple Classifiers System for Anomaly Detection in Credit Card Data With Unbalanced and Overlapped Classes

Contact Info

Product

Resources

About