Scalable Multi-Output Label Prediction: From Classifier Chains to Classifier Trellises

Abstract

Multi-output inference tasks, such as multi-label classification, have become increasingly important in recent years. A popular method for multi-label classification is classifier chains, in which the predictions of individual classifiers are cascaded along a chain, thus taking into account inter-label dependencies and improving the overall performance. Several varieties of classifier chain methods have been introduced, and many of them perform very competitively across a wide range of benchmark datasets. However, scalability limitations become apparent on larger datasets when modeling a fully-cascaded chain. In particular, the methods' strategies for discovering and modeling a good chain structure constitutes a mayor computational bottleneck. In this paper, we present the classifier trellis (CT) method for scalable multi-label classification. We compare CT with several recently proposed classifier chain methods to show that it occupies an important niche: it is highly competitive on standard multi-label problems, yet it can also scale up to thousands or even tens of thousands of labels.

(accepted in Pattern Recognition)

References in corpus (2)

• Estimating high-dimensional directed acyclic graphs with the PC-algorithm
• Efficient Monte Carlo Methods for Multi-Dimensional Learning with Classifier Chains

Cited by in corpus (7)

• Classifier Chains: A Review and Perspectives
• Many-Class Few-Shot Learning on Multi-Granularity Class Hierarchy
• Multi-label Methods for Prediction with Sequential Data
• A Confidence-based Acquisition Model for Self-supervised Active Learning and Label Correction
• Consistent Classification with Generalized Metrics
• mldr.resampling: Efficient Reference Implementations of Multilabel Resampling Algorithms
• Regular Partitions and Their Use in Structural Pattern Recognition