March 24, 2024Open Access

Toward Robustness in Multi-Label Classification: A Data Augmentation Strategy against Imbalance and Noise

View Full Paper

Ask AI

Bookmark

Authors

HSHwanjun SongKorea Advanced Institute of Science and Technology MKMinseok KimKwangwoon University JLJae-Gil LeeKorea Advanced Institute of Science and Technology

Discussion

Member takes

Implication

Benchmark evaluation demonstrates superior multi-label classification accuracy across three datasets, indicating improved robustness against label noise and class imbalance.

Key Points

Data augmentation with BalanceMix improves multi-label classification performance across three benchmark datasets, surpassing current state-of-the-art baseline methods.
Benchmark evaluation on three datasets demonstrates that dual samplers effectively generate diverse minority-augmented instances while refining noisy labels at label-wise granularity.
Robust optimization through categorization of noisy labels into clean, re-labeled, or ambiguous subsets provides a practical framework for handling imbalanced labels in training data.

Cite This Study

Song et al. (2024) studied this question.

synapsesocial.com/papers/68e72954b6db6435876a2fc7 https://doi.org/10.1609/aaai.v38i19.30157

View Full Paper

Ask AI

Bookmark

Implication

Benchmark evaluation demonstrates superior multi-label classification accuracy across three datasets, indicating improved robustness against label noise and class imbalance.

Toward Robustness in Multi-Label Classification: A Data Augmentation Strategy against Imbalance and Noise

Discussion

Member takes

Implication

Key Points

Cite This Study

Toward Robustness in Multi-Label Classification: A Data Augmentation Strategy against Imbalance and Noise

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider