Abstract:Conventional multi-label classification (MLC) methods assume that all samples are fully labeled and identically distributed. Unfortunately, this assumption is unrealistic in large-scale MLC data that has long-tailed (LT) distribution and partial labels (PL). To address the problem, we introduce a novel task, Partial labeling and Long-Tailed Multi-Label Classification (PLT-MLC), to jointly consider the above two imperfect learning environments. Not surprisingly, we find that most LT-MLC and PL-MLC approaches fail to solve the PLT-MLC, resulting in significant performance degradation on the two proposed PLT-MLC benchmarks. Therefore, we propose an end-to-end learning framework: \textbf{CO}rrection $\rightarrow$ \textbf{M}odificat\textbf{I}on $\rightarrow$ balan\textbf{C}e, abbreviated as \textbf{\method{}}. Our bootstrapping philosophy is to simultaneously correct the missing labels (Correction) with convinced prediction confidence over a class-aware threshold and to learn from these recall labels during training. We next propose a novel multi-focal modifier loss that simultaneously addresses head-tail imbalance and positive-negative imbalance to adaptively modify the attention to different samples (Modification) under the LT class distribution. In addition, we develop a balanced training strategy by distilling the model's learning effect from head and tail samples, and thus design a balanced classifier (Balance) conditioned on the head and tail learning effect to maintain stable performance for all samples. Our experimental study shows that the proposed \method{} significantly outperforms general MLC, LT-MLC and PL-MLC methods in terms of effectiveness and robustness on our newly created PLT-MLC datasets.

Long-Tailed Partial Label Learning by Head Classifier and Tail Classifier Cooperation

Long-Tailed Partial Label Learning Via Dynamic Rebalancing

Partial Label Learning with a Partner.

Complementary Classifier Induced Partial Label Learning

Learning in Imperfect Environment: Multi-Label Classification with Long-Tailed Distribution and Partial Labels

Partial label learning with heterogeneous domain adaptation

Can Label-Specific Features Help Partial-Label Learning?

Partial Multi-label Learning with Label and Feature Collaboration

Feature space and label space selection based on Error-correcting output codes for partial label learning

Partial Label Learning with Noisy Side Information

Large Margin Partial Label Machine

SoLar: Sinkhorn Label Refinery for Imbalanced Partial-Label Learning

Learning for Tail Label Data: A Label-Specific Feature Approach.

An Unbiased Risk Estimator for Partial Label Learning with Augmented Classes

Appeal: Allow Mislabeled Samples the Chance to be Rectified in Partial Label Learning

Partial Label Metric Learning Based on Statistical Inference

Exploiting Class Activation Value for Partial-Label Learning.

On Characterizing and Mitigating Imbalances in Multi-Instance Partial Label Learning

Partial-label Learning with Mixed Closed-set and Open-set Out-of-candidate Examples

Pseudo Labels for Imbalanced Multi-Label Learning

Three Heads Are Better Than One: Complementary Experts for Long-Tailed Semi-supervised Learning