Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2511.13759 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912715396612096 |
|---|---|
| author | Wang, Han Ji, Deyi Lu, Junyu Zhu, Lanyun Zhang, Hailong Wu, Haiyang Liu, Liqun Shu, Peng Lee, Roy Ka-Wei |
| author_facet | Wang, Han Ji, Deyi Lu, Junyu Zhu, Lanyun Zhang, Hailong Wu, Haiyang Liu, Liqun Shu, Peng Lee, Roy Ka-Wei |
| contents | Accurate detection of offensive content on social media demands high-quality labeled data; however, such data is often scarce due to the low prevalence of offensive instances and the high cost of manual annotation. To address this low-resource challenge, we propose a self-training framework that leverages abundant unlabeled data through collaborative pseudo-labeling. Starting with a lightweight classifier trained on limited labeled data, our method iteratively assigns pseudo-labels to unlabeled instances with the support of Multi-Agent Vision-Language Models (MA-VLMs). Un-labeled data on which the classifier and MA-VLMs agree are designated as the Agreed-Unknown set, while conflicting samples form the Disagreed-Unknown set. To enhance label reliability, MA-VLMs simulate dual perspectives, moderator and user, capturing both regulatory and subjective viewpoints. The classifier is optimized using a novel Positive-Negative-Unlabeled (PNU) loss, which jointly exploits labeled, Agreed-Unknown, and Disagreed-Unknown data while mitigating pseudo-label noise. Experiments on benchmark datasets demonstrate that our framework substantially outperforms baselines under limited supervision and approaches the performance of large-scale models |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_13759 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Multi-Agent VLMs Guided Self-Training with PNU Loss for Low-Resource Offensive Content Detection Wang, Han Ji, Deyi Lu, Junyu Zhu, Lanyun Zhang, Hailong Wu, Haiyang Liu, Liqun Shu, Peng Lee, Roy Ka-Wei Machine Learning Artificial Intelligence I.2.10 Accurate detection of offensive content on social media demands high-quality labeled data; however, such data is often scarce due to the low prevalence of offensive instances and the high cost of manual annotation. To address this low-resource challenge, we propose a self-training framework that leverages abundant unlabeled data through collaborative pseudo-labeling. Starting with a lightweight classifier trained on limited labeled data, our method iteratively assigns pseudo-labels to unlabeled instances with the support of Multi-Agent Vision-Language Models (MA-VLMs). Un-labeled data on which the classifier and MA-VLMs agree are designated as the Agreed-Unknown set, while conflicting samples form the Disagreed-Unknown set. To enhance label reliability, MA-VLMs simulate dual perspectives, moderator and user, capturing both regulatory and subjective viewpoints. The classifier is optimized using a novel Positive-Negative-Unlabeled (PNU) loss, which jointly exploits labeled, Agreed-Unknown, and Disagreed-Unknown data while mitigating pseudo-label noise. Experiments on benchmark datasets demonstrate that our framework substantially outperforms baselines under limited supervision and approaches the performance of large-scale models |
| title | Multi-Agent VLMs Guided Self-Training with PNU Loss for Low-Resource Offensive Content Detection |
| topic | Machine Learning Artificial Intelligence I.2.10 |
| url | https://arxiv.org/abs/2511.13759 |