AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

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

Han Wang, Deyi Ji, Junyu Lu, Lanyun Zhu, Hailong Zhang, Haiyang Wu, Liqun Liu, Peng Shu, Roy Ka-Wei Lee

AAAI 2026 · Label-efficient and robust learning › Semi-supervised learning · approach: Confidence-based pseudo-label selection

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add self-training with CLIP + Qwen72B pseudo-labels to the CLIP classifierCLIP-Large, n=100 · FHMmacro-F1: 59.24 -> 72.68Improved
add self-training with CLIP + Qwen72B pseudo-labels to the CLIP classifierCLIP-Large, n=100 · Sent140macro-F1: 64.22 -> 77.11Improved
verify classifier pseudo-labels with MA-VLMs (CLIP + Qwen72B) instead of CLIP-only labels (ablation)SelfTrain CLIP, n=100 · MAMI Task Amacro-F1: 67.03 -> 73.49Improved
verify classifier pseudo-labels with MA-VLMs on HSOL (ablation)SelfTrain CLIP, n=100 · HSOLmacro-F1: 86.48 -> 86.69No clear effect
use MA-VLMs moderator/user prompt instead of zero-shot (ablation)Qwen2.5-VL-72B · MAMI Task Amacro-F1: 79.17 -> 81.64Improved
use MA-VLMs prompt instead of zero-shot (ablation)Qwen2.5-VL-72B · FHMmacro-F1: 74.46 -> 74.62No clear effect
add few-shot examples to the VLM prompt (ablation)Qwen2.5-VL-72B · FHMmacro-F1: 74.46 -> 71.09Hurt
add chain-of-thought to the VLM prompt (ablation)Qwen2.5-VL-72B · FHM and MAMImacro-F1: 74.46 -> 74.43 (FHM)No clear effect
negative PNU weight gamma = -0.1 (NU learning) (ablation)SelfTrain CLIP, n=100 · MAMI Task Amacro-F1: 73.49 -> 59.98Hurt
set PNU gamma = 0.1 instead of 0.0 (add PU term) on MAMI (ablation)SelfTrain CLIP, n=100 · MAMI Task Amacro-F1: 68.84 -> 73.49Improved
PNU gamma = 0.1 instead of 0.0 on FHM (ablation)SelfTrain CLIP, n=100 · FHMmacro-F1: 72.68 -> 71.50Hurt
increase labels from n=100 to n=250 for SupOnly Qwen7B vs SelfTrain CLIP (ablation)Qwen7B vs SelfTrain CLIP · FHMmacro-F1: SelfTrain 72.97 vs Qwen7B 75.88 at n=250Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy