Deep Partial Multi-Label Learning with Graph Disambiguation
This addresses the challenge of handling noisy label sets in multi-label classification, which is incremental as it extends existing graph-based methods to deep models.
The paper tackles the problem of partial multi-label learning, where data has candidate label sets containing both true and false labels, by proposing a deep model with graph disambiguation that recovers label confidences and dependencies, achieving significantly superior results compared to state-of-the-art methods on synthetic and real-world datasets.
In partial multi-label learning (PML), each data example is equipped with a candidate label set, which consists of multiple ground-truth labels and other false-positive labels. Recently, graph-based methods, which demonstrate a good ability to estimate accurate confidence scores from candidate labels, have been prevalent to deal with PML problems. However, we observe that existing graph-based PML methods typically adopt linear multi-label classifiers and thus fail to achieve superior performance. In this work, we attempt to remove several obstacles for extending them to deep models and propose a novel deep Partial multi-Label model with grAph-disambIguatioN (PLAIN). Specifically, we introduce the instance-level and label-level similarities to recover label confidences as well as exploit label dependencies. At each training epoch, labels are propagated on the instance and label graphs to produce relatively accurate pseudo-labels; then, we train the deep model to fit the numerical labels. Moreover, we provide a careful analysis of the risk functions to guarantee the robustness of the proposed model. Extensive experiments on various synthetic datasets and three real-world PML datasets demonstrate that PLAIN achieves significantly superior results to state-of-the-art methods.