Dealing with Data Difficulty Factors While Learning from Imbalanced Data

Hypergraph-based importance assessment for binary classification data

We present a novel hypergraph-based framework enabling an assessment of the importance of binary classification data elements. Specifically, we apply the hypergraph model to rate data samples’ and categorical feature values’ relevance to classification labels. The proposed Hypergraph-based Importance ratings are theoretically grounded on the hypergraph cut conductance minimization concept. As a result of using hypergraph representation, which is a lossless representation from the perspective of higher-order relationships in data, our approach allows for more precise exploitation of the information on feature and sample coincidences. The solution was tested using two scenarios: undersampling for imbalanced classification data and feature selection. The experimentation results have proven the good quality of the new approach when compared with other state-of-the-art and baseline methods for both scenarios measured using the average precision evaluation metric.

Double decoupled network for imbalanced obstetric intelligent diagnosis

Electronic Medical Record (EMR) is the data basis of intelligent diagnosis. The diagnosis results of an EMR are multi-disease, including normal diagnosis, pathological diagnosis and complications, so intelligent diagnosis can be treated as multi-label classification problem. The distribution of diagnostic results in EMRs is imbalanced. And the diagnostic results in one EMR have a high coupling degree. The traditional rebalancing methods does not function effectively on highly coupled imbalanced datasets. This paper proposes Double Decoupled Network (DDN) based intelligent diagnosis model, which decouples representation learning and classifier learning. In the representation learning stage, Convolutional Neural Networks (CNN) is used to learn the original features of the data. In the classifier learning stage, a Decoupled and Rebalancing highly Imbalanced Labels (DRIL) algorithm is proposed to decouple the highly coupled diagnostic results and rebalance the datasets, and then the balanced datasets is used to train the classifier. This paper evaluates the proposed DDN using Chinese Obstetric EMR (COEMR) datasets, and verifies the effectiveness and universality of the model on two benchmark multi-label text classification datasets: Arxiv Academic Papers Datasets (AAPD) and Reuters Corpus1 (RCV1). Demonstrating the effectiveness of the proposed methods is an imbalanced obstetric EMRs. The accuracy of DDN model on COEMR, AAPD and RCV1 datasets is 84.17, 86.35 and 93.87% respectively, which is higher than the current optimal experimental results.

High-Dimensional Unbalanced Binary Classification by Genetic Programming with Multi-Criterion Fitness Evaluation and Selection

High-dimensional unbalanced classification is challenging because of the joint effects of high dimensionality and class imbalance. Genetic programming (GP) has the potential benefits for use in high-dimensional classification due to its built-in capability to select informative features. However, once data are not evenly distributed, GP tends to develop biased classifiers which achieve a high accuracy on the majority class but a low accuracy on the minority class. Unfortunately, the minority class is often at least as important as the majority class. It is of importance to investigate how GP can be effectively utilized for high-dimensional unbalanced classification. In this article, to address the performance bias issue of GP, a new two-criterion fitness function is developed, which considers two criteria, that is, the approximation of area under the curve (AUC) and the classification clarity (i.e., how well a program can separate two classes). The obtained values on the two criteria are combined in pairs, instead of summing them together. Furthermore, this article designs a three-criterion tournament selection to effectively identify and select good programs to be used by genetic operators for generating offspring during the evolutionary learning process. The experimental results show that the proposed method achieves better classification performance than other compared methods.

RB-CCR: Radial-Based Combined Cleaning and Resampling algorithm for imbalanced data classification

Real-world classification domains, such as medicine, health and safety, and finance, often exhibit imbalanced class priors and have asynchronous misclassification costs. In such cases, the classification model must achieve a high recall without significantly impacting precision. Resampling the training data is the standard approach to improving classification performance on imbalanced binary data. However, the state-of-the-art methods ignore the local joint distribution of the data or correct it as a post-processing step. This can causes sub-optimal shifts in the training distribution, particularly when the target data distribution is complex. In this paper, we propose Radial-Based Combined Cleaning and Resampling (RB-CCR). RB-CCR utilizes the concept of class potential to refine the energy-based resampling approach of CCR. In particular, RB-CCR exploits the class potential to accurately locate sub-regions of the data-space for synthetic oversampling. The category sub-region for oversampling can be specified as an input parameter to meet domain-specific needs or be automatically selected via cross-validation. Our $$5\times 2$$ 5 × 2 cross-validated results on 57 benchmark binary datasets with 9 classifiers show that RB-CCR achieves a better precision-recall trade-off than CCR and generally out-performs the state-of-the-art resampling methods in terms of AUC and G-mean.

The impact of data difficulty factors on classification of imbalanced and concept drifting data streams

Class imbalance introduces additional challenges when learning classifiers from concept drifting data streams. Most existing work focuses on designing new algorithms for dealing with the global imbalance ratio and does not consider other data complexities. Independent research on static imbalanced data has highlighted the influential role of local data difficulty factors such as minority class decomposition and presence of unsafe types of examples. Despite often being present in real-world data, the interactions between concept drifts and local data difficulty factors have not been investigated in concept drifting data streams yet. We thoroughly study the impact of such interactions on drifting imbalanced streams. For this purpose, we put forward a new categorization of concept drifts for class imbalanced problems. Through comprehensive experiments with synthetic and real data streams, we study the influence of concept drifts, global class imbalance, local data difficulty factors, and their combinations, on predictions of representative online classifiers. Experimental results reveal the high influence of new considered factors and their local drifts, as well as differences in existing classifiers’ reactions to such factors. Combinations of multiple factors are the most challenging for classifiers. Although existing classifiers are partially capable of coping with global class imbalance, new approaches are needed to address challenges posed by imbalanced data streams.

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

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.

Using Item Response Theory for Explainable Machine Learning in Predicting Mortality in the Intensive Care Unit: Case-Based Approach

BACKGROUND

Supervised machine learning (ML) is being featured in the health care literature with study results frequently reported using metrics such as accuracy, sensitivity, specificity, recall, or F1 score. Although each metric provides a different perspective on the performance, they remain to be overall measures for the whole sample, discounting the uniqueness of each case or patient. Intuitively, we know that all cases are not equal, but the present evaluative approaches do not take case difficulty into account.

OBJECTIVE

A more case-based, comprehensive approach is warranted to assess supervised ML outcomes and forms the rationale for this study. This study aims to demonstrate how the item response theory (IRT) can be used to stratify the data based on how difficult each case is to classify, independent of the outcome measure of interest (eg, accuracy). This stratification allows the evaluation of ML classifiers to take the form of a distribution rather than a single scalar value.

METHODS

Two large, public intensive care unit data sets, Medical Information Mart for Intensive Care III and electronic intensive care unit, were used to showcase this method in predicting mortality. For each data set, a balanced sample (n=8078 and n=21,940, respectively) and an imbalanced sample (n=12,117 and n=32,910, respectively) were drawn. A 2-parameter logistic model was used to provide scores for each case. Several ML algorithms were used in the demonstration to classify cases based on their health-related features: logistic regression, linear discriminant analysis, K-nearest neighbors, decision tree, naive Bayes, and a neural network. Generalized linear mixed model analyses were used to assess the effects of case difficulty strata, ML algorithm, and the interaction between them in predicting accuracy.

RESULTS

The results showed significant effects (P<.001) for case difficulty strata, ML algorithm, and their interaction in predicting accuracy and illustrated that all classifiers performed better with easier-to-classify cases and that overall the neural network performed best. Significant interactions suggest that cases that fall in the most arduous strata should be handled by logistic regression, linear discriminant analysis, decision tree, or neural network but not by naive Bayes or K-nearest neighbors. Conventional metrics for ML classification have been reported for methodological comparison.

CONCLUSIONS

This demonstration shows that using the IRT is a viable method for understanding the data that are provided to ML algorithms, independent of outcome measures, and highlights how well classifiers differentiate cases of varying difficulty. This method explains which features are indicative of healthy states and why. It enables end users to tailor the classifier that is appropriate to the difficulty level of the patient for personalized medicine.

Tackling the Problem of Class Imbalance in Multi-class Sentiment Classification: An Experimental Study

Sentiment classification is an important task which gained extensive attention both in academia and in industry. Many issues related to this task such as handling of negation or of sarcastic utterances were analyzed and accordingly addressed in previous works. However, the issue of class imbalance which often compromises the prediction capabilities of learning algorithms was scarcely studied. In this work, we aim to bridge the gap between imbalanced learning and sentiment analysis. An experimental study including twelve imbalanced learning preprocessing methods, four feature representations, and a dozen of datasets, is carried out in order to analyze the usefulness of imbalanced learning methods for sentiment classification. Moreover, the data difficulty factors — commonly studied in imbalanced learning — are investigated on sentiment corpora to evaluate the impact of class imbalance.

A Pareto-based Ensemble with Feature and Instance Selection for Learning from Multi-Class Imbalanced Datasets

Imbalanced classification is related to those problems that have an uneven distribution among classes. In addition to the former, when instances are located into the overlapped areas, the correct modeling of the problem becomes harder. Current solutions for both issues are often focused on the binary case study, as multi-class datasets require an additional effort to be addressed. In this research, we overcome these problems by carrying out a combination between feature and instance selections. Feature selection will allow simplifying the overlapping areas easing the generation of rules to distinguish among the classes. Selection of instances from all classes will address the imbalance itself by finding the most appropriate class distribution for the learning task, as well as possibly removing noise and difficult borderline examples. For the sake of obtaining an optimal joint set of features and instances, we embedded the searching for both parameters in a Multi-Objective Evolutionary Algorithm, using the C4.5 decision tree as baseline classifier in this wrapper approach. The multi-objective scheme allows taking a double advantage: the search space becomes broader, and we may provide a set of different solutions in order to build an ensemble of classifiers. This proposal has been contrasted versus several state-of-the-art solutions on imbalanced classification showing excellent results in both binary and multi-class problems.