The importance of the label hierarchy in hierarchical multi-label classification

Abstract

We address the task of hierarchical multi-label classification (HMC). HMC is a task of structured output prediction where the classes are organized into a hierarchy and an instance may belong to multiple classes. In many problems, such as gene function prediction or prediction of ecological community structure, classes inherently follow these constraints. The potential for application of HMC was recognized by many researchers and several such methods were proposed and demonstrated to achieve good predictive performances in the past. However, there is no clear understanding when is favorable to consider such relationships (hierarchical and multi-label) among classes, and when this presents unnecessary burden for classification methods. To this end, we perform a detailed comparative study over 8 datasets that have HMC properties. We investigate two important influences in HMC: the multiple labels per example and the information about the hierarchy. More specifically, we consider four machine learning tasks: multi-label classification, hierarchical multi-label classification, single-label classification and hierarchical single-label classification. To construct the predictive models, we use predictive clustering trees (a generalized form of decision trees), which are able to tackle each of the modelling tasks listed. Moreover, we investigate whether the influence of the hierarchy and the multiple labels carries over for ensemble models. For each of the tasks, we construct a single tree and two ensembles (random forest and bagging). The results reveal that the hierarchy and the multiple labels do help to obtain a better single tree model, while this is not preserved for the ensemble models.

Visually-Enabled Active Deep Learning for (Geo) Text and Image Classification: A Review

“…Both binary and multi-class classifications are "single-label" methods (thus, binary/multi-class classifications is also called single-label classification in the literature [205]), where each instance is only associated with a single class label (see Figure A1a,b for an illustration). By contrast, multi-label classification (also multilabel classification) produces a labeled data set where each instance is associated with a vector of output values [190,[206][207][208][209], instead of only one value.…”

“…Hierarchical classification combined with single-label classification (Appendix A.4.3) are called hierarchical single-label classification (HSC) in the literature [205]. Vailaya et al [215] provided an early example of hierarchical classification combined with binary classification (Appendix A.4.1).…”

A Weighted Voting Ensemble Self-Labeled Algorithm for the Detection of Lung Abnormalities from X-Rays

“…In our future work, we intend to pursue extensive empirical experiments to compare the proposed WvEnSL with other algorithms belonging to different SSL classes, and evaluate its performance using various component self-labeled algorithms and base learners. Furthermore, since our preliminary numerical experiments are quite encouraging, our next step is to explore the performance of the proposed algorithm on imbalanced datasets [39,40] and incorporate our proposed methodology for multi-target problems [41][42][43]. Additionally, another interesting aspect is the use of other component classifiers in the ensemble and enhance our proposed framework with more sophisticated and theoretically sound criteria for the development of an advanced weighted voting strategy.…”

Labelling strategies for hierarchical multi-label classification techniques

“…Numbers reported in the HMC literature range from a single global threshold used for all datasets [13] to a single threshold per dataset [28,29,30,31]. Although it has been done in non-hierarchical multi-label classification, to our knowledge the selection of a separate threshold per class has not been performed in the HMC context.…”