让多模态大模型更懂长尾数据分布，提升极端值预测能力。
Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression
- 用批次对比奖励机制，让模型学全量数据的分布关系。
- 在长尾回归任务中显著降低均值偏差，尤其在少样本时提升明显。
- 无需改架构，可直接接入现有模型，适合数据不均衡场景。
多模态大语言模型在长尾目标分布下的数值回归任务中表现不佳。基于标记级别的监督微调（SFT）和点级回归奖励会使学习偏向高密度区域，导致回归均值化现象，尾部性能差。我们发现现有训练范式缺乏跨样本的相对监督是关键瓶颈。为此，提出一种基于组相对策略优化的分布感知强化学习框架，通过基于一致性相关系数的奖励引入批次级对比监督，使预测与真实分布的关联性、尺度和均值对齐。该框架即插即用，无需架构修改。在统一的长尾回归基准测试集上，相比SFT及现有MLLM回归方法，实现一致改进，尤其在中等和少样本情形下增益显著。
原文摘要 · Abstract (English)
Multimodal large language models (MLLMs) struggle with numerical regression under long-tailed target distributions. Token-level supervised fine-tuning (SFT) and point-wise regression rewards bias learning toward high-density regions, leading to regression-to-the-mean behavior and poor tail performance. We identify the lack of cross-sample relational supervision as a key limitation of existing MLLM training paradigms. To address it, we propose a distribution-aware reinforcement learning framework based on Group Relative Policy Optimization, which introduces batch-level comparison-based supervision via the Concordance Correlation Coefficient-based reward to align predicted and ground-truth distributions in terms of correlation, scale, and mean. The framework is plug-and-play, requiring no architectural modification. Experiments on a unified suite of long-tailed regression benchmarks show consistent improvements over SFT and existing MLLM regression methods, with particularly strong gains in medium- and few-shot regimes.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。