# HarmTrace: Anchor-Calibrated Decoupled Optimization for Fine-Grained Target Identification in Harmful Memes

arXiv: https://arxiv.org/abs/2608.16622 · HTML: https://mlsanity.com/paper/2608.16622/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 18 of its 41 references that are in the map.
- First posted on arXiv 17 Aug 2026.

## Builds on (most cited first, 18 of 18)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://mlsanity.com/paper/2508.18265/) (2025)
- [Gemma: Open Models Based on Gemini Research and Technology](https://mlsanity.com/paper/2403.08295/) (2024)
- [Kimi K2.5: Visual Agentic Intelligence](https://mlsanity.com/paper/2602.02276/) (2026)
- [Decoding the Underlying Meaning of Multimodal Hateful Memes](https://mlsanity.com/paper/2305.17678/) (2023)
- [GLM-5V-Turbo: Toward a Native Foundation Model for Multimodal Agents](https://mlsanity.com/paper/2604.26752/) (2026)
- [NGRPO: Negative-enhanced Group Relative Policy Optimization](https://mlsanity.com/paper/2509.18851/) (2025)
- [Demystifying Hateful Content: Leveraging Large Multimodal Models for Hateful Meme Detection with Explainable Decisions](https://mlsanity.com/paper/2502.11073/) (2025)
- [PAPO: Stabilizing Rubric Integration Training via Decoupled Advantage Normalization](https://mlsanity.com/paper/2603.26535/) (2026)
- [MemeIntel: Explainable Detection of Propagandistic and Hateful Memes](https://mlsanity.com/paper/2502.16612/) (2025)
- [ExPO-HM: Learning to Explain-then-Detect for Hateful Meme Detection](https://mlsanity.com/paper/2510.08630/) (2025)
- [MemeLens: Multilingual Multitask VLMs for Memes](https://mlsanity.com/paper/2601.12539/) (2026)
- [From Recognition to Reasoning: Advancing Multimodal Harmful Meme Detection via Chain-of-Thought Alignment](https://mlsanity.com/paper/2506.18919/) (2025)
- [Can Thinking Models Think to Detect Hateful Memes?](https://mlsanity.com/paper/2603.01225/) (2026)

## Abstract (arXiv)

Multimodal harmful meme detection is typically formulated as image--text harmfulness classification. A model may correctly predict harmfulness while misidentifying the attacked target or its supporting evidence. We therefore extend harmful meme detection with fine-grained target identification, asking what type of target is attacked, who is targeted, and where the target appears in the meme. The model predicts harmfulness for every meme and, for harmful memes, outputs the target category, target entity, textual mention, and visual region. To support this task, we introduce Meme3W, which unifies multiple public harmful meme datasets and provides human-verified annotations for harmful instances. We further introduce Joint Record Accuracy (JRA), a strict record-level metric requiring the harmfulness label and all target-identification fields to be jointly correct. Experiments with representative multimodal large language models reveal a substantial gap between harmfulness accuracy and JRA. To narrow this gap, we propose HarmTrace, an anchor-calibrated decoupled optimization framework. HarmTrace strengthens target-entity supervision through entity-aware supervised fine-tuning. It then applies Conditional Target-identification Policy Optimization (CTPO) to decouple harmfulness and target-identification advantages, restricting target-identification optimization to label-correct responses for harmful examples. CTPO uses a Virtual Positive Anchor (VPA) as a fully correct reference for target-identification advantage normalization. HarmTrace improves both JRA and harmfulness accuracy across the evaluated backbones, with JRA on the Qwen3-VL-8B backbone increasing from 17.58% to 52.51%. Our code is publicly available at https://github.com/llly1234/HarmTrace-for-Harmful-Memes.