一作ACL 2026Findings· Findings of ACL
DR - HM : Distill-then-Reinforce Training with Cognition-Aware Data Synthesis for Harmful Meme Detection
Zihan Cheng · Jianxiang Ma · Xiaocui Yang · Peidong Wang · Wen Zhang · Shi Feng · Daling Wang · Yifei Zhang · Mingfu Zhang
摘要
Harmful memes convey offensive intent through implicit associations between visual symbols and text, requiring a broad understanding of cultural stereotypes and visual metaphors. Small-scale Multimodal Large Language Models (MLLMs) often lack the knowledge required to identify such implicit hate,...