# Multi-Agent VLMs Guided Self-Training with PNU Loss for Low-Resource Offensive Content Detection

arXiv: https://arxiv.org/abs/2511.13759 · HTML: https://mlsanity.com/paper/2511.13759/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Social simulation and multi-agent debate, one of 16 directions in LLM agents and tool use.
- Builds on 11 of its 55 references that are in the map.
- First posted on arXiv 14 Nov 2025.

## Placed in

- LLM agents and tool use / [Social simulation and multi-agent debate](https://mlsanity.com/agents/social-simulation-and-multi-agent-debate/)
- Synthetic data and self-improvement for LLMs / [Self-training for few-shot NLP](https://mlsanity.com/synth/self-training-for-few-shot-nlp/)

## Builds on (most cited first, 11 of 11)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [Large Language Model based Multi-Agents: A Survey of Progress and Challenges](https://mlsanity.com/paper/2402.01680/) (2024)
- [Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes](https://mlsanity.com/paper/2305.02301/) (2023)
- [To CoT or not to CoT? Chain-of-thought helps mainly on math and symbolic reasoning](https://mlsanity.com/paper/2409.12183/) (2024)
- [Detecting Hate Speech with GPT-3](https://mlsanity.com/paper/2103.12407/) (2021)
- [Evaluating GPT-3 Generated Explanations for Hateful Content Moderation](https://mlsanity.com/paper/2305.17680/) (2023)
- [Improving Hateful Meme Detection through Retrieval-Guided Contrastive Learning](https://mlsanity.com/paper/2311.08110/) (2023)
- [Towards Interpretable Hate Speech Detection using Large Language Model-extracted Rationales](https://mlsanity.com/paper/2403.12403/) (2024)
- [Combating Adversarial Attacks with Multi-Agent Debate](https://mlsanity.com/paper/2401.05998/) (2024)
- [DST-Det: Simple Dynamic Self-Training for Open-Vocabulary Object Detection](https://mlsanity.com/paper/2310.01393/) (2023)

## Built on by (most cited first, 2 of 2)

- [RAVEN++: Pinpointing Fine-Grained Violations in Advertisement Videos with Active Reinforcement Reasoning](https://mlsanity.com/paper/2511.19168/) (2025)
- [Generalized Distribution-Free Semi-Supervised Learning with Risk Rewrite](https://mlsanity.com/paper/2607.11947/) (2026)

## Abstract (arXiv)

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