ExPO-HM: Learning to Explain-then-Detect for Hateful Meme Detection

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Mei, Jingbiao, Sun, Mingsheng, Chen, Jinghong, Qin, Pengda, Li, Yuhong, Chen, Da, Byrne, Bill
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866918362586546176
author Mei, Jingbiao
Sun, Mingsheng
Chen, Jinghong
Qin, Pengda
Li, Yuhong
Chen, Da
Byrne, Bill
author_facet Mei, Jingbiao
Sun, Mingsheng
Chen, Jinghong
Qin, Pengda
Li, Yuhong
Chen, Da
Byrne, Bill
contents Hateful memes have emerged as a particularly challenging form of online abuse, motivating the development of automated detection systems. Most prior approaches rely on direct detection, producing only binary predictions. Such models fail to provide the context and explanations that real-world moderation requires. Recent Explain-then-Detect approaches, using Chain-of-Thought prompting or LMM agents, perform worse than simple SFT baselines, and even advanced post-training methods such as GRPO fail to close the gap. Our analysis identifies two key issues of such systems: important policy-relevant cues such as targets and attack types are not hypothesized by the model as a likely explanation; and the binary reward signal is insufficient to guide reasoning. To address these challenges, we propose ExPO-HM (Explain-then-Detect Policy Optimization for Hateful Memes), inspired by the training and evaluation process of human annotators. ExPO-HM combines SFT warmup, GRPO with curriculum learning, and Conditional Decision Entropy (CDE) as both metric and reward for reasoning quality. Across three hateful meme benchmarks, ExPO-HM achieves state-of-the-art performance on binary detection, fine-grained classification, and reasoning quality, with up to 15\% and 17\% F1 improvement over the GRPO and DPO baselines, respectively. By moving hateful meme detection from simple binary alarms to explanation-driven detection, ExPO-HM provides accurate, interpretable, and actionable moderation support. Code available at https://github.com/JingbiaoMei/ExPO-HM
format Preprint
id arxiv_https___arxiv_org_abs_2510_08630
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ExPO-HM: Learning to Explain-then-Detect for Hateful Meme Detection
Mei, Jingbiao
Sun, Mingsheng
Chen, Jinghong
Qin, Pengda
Li, Yuhong
Chen, Da
Byrne, Bill
Computation and Language
Hateful memes have emerged as a particularly challenging form of online abuse, motivating the development of automated detection systems. Most prior approaches rely on direct detection, producing only binary predictions. Such models fail to provide the context and explanations that real-world moderation requires. Recent Explain-then-Detect approaches, using Chain-of-Thought prompting or LMM agents, perform worse than simple SFT baselines, and even advanced post-training methods such as GRPO fail to close the gap. Our analysis identifies two key issues of such systems: important policy-relevant cues such as targets and attack types are not hypothesized by the model as a likely explanation; and the binary reward signal is insufficient to guide reasoning. To address these challenges, we propose ExPO-HM (Explain-then-Detect Policy Optimization for Hateful Memes), inspired by the training and evaluation process of human annotators. ExPO-HM combines SFT warmup, GRPO with curriculum learning, and Conditional Decision Entropy (CDE) as both metric and reward for reasoning quality. Across three hateful meme benchmarks, ExPO-HM achieves state-of-the-art performance on binary detection, fine-grained classification, and reasoning quality, with up to 15\% and 17\% F1 improvement over the GRPO and DPO baselines, respectively. By moving hateful meme detection from simple binary alarms to explanation-driven detection, ExPO-HM provides accurate, interpretable, and actionable moderation support. Code available at https://github.com/JingbiaoMei/ExPO-HM
title ExPO-HM: Learning to Explain-then-Detect for Hateful Meme Detection
topic Computation and Language
url https://arxiv.org/abs/2510.08630