Abstract:Partial label learning (PLL) induces a multi-class classifier from training examples each associated with a set of candidate labels, among which only one is valid. The formation of real-world data typically arises from heterogeneous entanglement of series latent explanatory factors, which are considered intrinsic properties for discriminating between different patterns. Though learning disentangled representation is expected to facilitate label disambiguation for partial-label (PL) examples, few existing works were dedicated to addressing this issue. In this paper, we make the first attempt towards disentangled PLL and propose a novel approach named TERIAL, which makes predictions according to derived disentangled representation of instances and label embeddings. The TERIAL approach formulates the PL examples as an undirected bipartite graph where instances are only connected with their candidate labels, and employs a tailored neighborhood routing mechanism to yield disentangled representation of nodes in the graph. Specifically, the proposed routing mechanism progressively infers the explanatory factors that contribute to the edge between adjacent nodes and augments the representation of the central node with factor-aware embedding information propagated from specific neighbors simultaneously via iteratively analyzing the promising subspace clusters formed by the node and its neighbors. The estimated labeling confidence matrix is also introduced to accommodate unreliable links owing to the inherent ambiguity of PLL. Moreover, we theoretically prove that the neighborhood routing mechanism will converge to the point estimate that maximizes the marginal likelihood of observed PL training examples. Comprehensive experiments over various datasets demonstrate that our approach outperforms the state-of-the-art counterparts.

Self-distillation and self-supervision for partial label learning

Self-Paced Method for Transfer Partial Label Learning

Partial Label Learning with a Partner.

Semi-supervised Partial Label Learning Algorithm Via Reliable Label Propagation

Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning

A Self-Paced Regularization Framework for Partial-Label Learning

Partial Label Learning Via Self-Paced Curriculum Strategy

A Self-Ensemble Approach for Partial Multi-Label Learning

Unreliable Partial Label Learning with Recursive Separation

Partial Label Learning with Self-Guided Retraining.

Understanding Self-Distillation and Partial Label Learning in Multi-Class Classification with Label Noise

Partial Label Learning with Semantic Label Representations

Disentangled Partial Label Learning

Controller-Guided Partial Label Consistency Regularization with Unlabeled Data

Can Label-Specific Features Help Partial-Label Learning?

Partial label learning with heterogeneous domain adaptation

Partial Label Learning Via Identifying Outlier Features

Label Confidence Based Ensemble Partial Label Learning

Partial Label Learning with Unlabeled Data.

Fine-Grained Partial Label Learning

Solving Partial Label Learning Problem with Multi-Agent Reinforcement Learning