Abstract:In multilabel classification, strong label dependence is present for exploiting, particularly for word-to-word dependence defined by semantic labels. In such a situation, we develop a collaborative-learning framework to predict class labels based on label-predictor pairs and label-only data. For example, in image categorization and recognition, language expressions describe the content of an image together with a large number of words and phrases without associated images. This article proposes a new loss quantifying partial correctness for false positive and negative misclassifications due to label similarities. Given this loss, we develop the Bayes rule to capture label dependence by nonlinear classification. On this ground, we introduce a weighted random forest classifier for complete data and a stacking scheme for leveraging additional labels to enhance the performance of supervised learning based on label-predictor pairs. Importantly, we decompose multilabel classification into a sequence of independent learning tasks, based on which the computational complexity of our classifier becomes linear in the size of labels. Compared to existing classifiers without label-only data, the proposed classifier enjoys the computational benefit while enabling the detection of novel labels absent from training by exploring label dependence and leveraging label-only data for higher accuracy. Theoretically, we show that the proposed method reconstructs the Bayes performance consistently, achieving the desired learning accuracy. Numerically, we demonstrate that the proposed method compares favorably in terms of the proposed and Hamming losses against binary relevance and a regularized Ising classifier modeling conditional label dependence. Indeed, leveraging additional labels tends to improve the supervised performance, especially when the training sample is not very large, as in semisupervised learning. Finally, we demonstrate the utility of the proposed approach on the Microsoft COCO object detection challenge, PASCAL visual object classes challenge 2007, and Mediamill benchmark.

A Formalization of Multilabel Classification in Terms of Lattice Theory and Information Theory: Concerning Datasets

Distributed Information-Theoretic Semisupervised Learning for Multilabel Classification

Label Disentanglement in Partition-based Extreme Multilabel Classification

Information-Theoretic Measures on Lattices for High-Order Interactions

Semi-Supervised Dual Relation Learning for Multi-Label Classification

Collaborative Multilabel Classification

Simulation for Lattice-Valued Doubly Labeled Transition Systems

Scalable Label Distribution Learning for Multi-Label Classification

A thorough experimental comparison of multilabel methods for classification performance

A Survey on Incomplete Multi-label Learning: Recent Advances and Future Trends

On the Persistence of Multilabel Learning, Its Recent Trends, and Its Open Issues

Hyperbolic Interaction Model For Hierarchical Multi-Label Classification

Data Mining Via Lattice Machine

Multi-label Classification using Labels as Hidden Nodes

A Label Information Aware Model for Multi-label Text Classification

A General Theory of Concept Lattice (I): Emergence of General Concept Lattice

Information-Theoretic Foundations for Machine Learning

Lattice Theory in Multi-Agent Systems

Research of multi-label text classification based on label attention and correlation networks

Complementary to Multiple Labels: A Correlation-Aware Correction Approach

Learning Disentangled Label Representations for Multi-label Classification