FLUID：通过令牌蒸馏实现专家专业化的流-潜在统一集成用于多模态学习
FLUID: Flow-Latent Unified Integration via Token Distillation for Expert Specialization in Multimodal Learning
AI总结：
本文提出FLUID框架，通过Q变换令牌蒸馏、两阶段融合和负载均衡专家混合模型，提升多模态分类的鲁棒性与可扩展性，在GLAMI-1M上达到91%准确率。
摘要：
多模态分类需要稳健地整合视觉和文本信号，然而常见的融合策略较为脆弱，容易受到模态特定噪声的影响。本文提出了FLUID——通过令牌蒸馏实现专家专业化的流-潜在统一集成，这是一个有原则的令牌级流程，可提高跨模态鲁棒性和可扩展性。FLUID包含三个核心要素：（1）Q变换，即可学习的查询令牌，用于从模态特定骨干网络中蒸馏并保留显著的令牌级特征；（2）两阶段融合方案，通过对比对齐强化跨模态一致性，然后通过门控机制和Q瓶颈执行自适应、任务感知的融合，该瓶颈选择性地压缩信息以供下游推理使用；（3）在预测阶段采用轻量级、负载均衡的专家混合模型，实现对多样化语义模式的高效专业化。大量实验表明，FLUID在GLAMI-1M基准上达到了91%的准确率，显著优于先前基线，并对标签噪声、长尾类别不平衡和语义异质性表现出强大的鲁棒性。有针对性的消融研究证实了所提组件各自及协同的益处，使FLUID成为多模态产品分类的可扩展、抗噪声解决方案。
英文摘要：
Multimodal classification requires robust integration of visual and textual signals, yet common fusion strategies are brittle and vulnerable to modality-specific noise. In this paper, we present \textsc{FLUID}-Flow-Latent Unified Integration via Token Distillation for Expert Specialization, a principled token-level pipeline that improves cross-modal robustness and scalability. \textsc{FLUID} contributes three core elements: (1) \emph{Q-transforms}, learnable query tokens that distill and retain salient token-level features from modality-specific backbones; (2) a two-stage fusion scheme that enforces cross-modal consistency via contrastive alignment and then performs adaptive, task-aware fusion through a gating mechanism and a \emph{Q-bottleneck} that selectively compresses information for downstream reasoning; and (3) a lightweight, load-balanced Mixture-of-Experts at prediction time that enables efficient specialization to diverse semantic patterns. Extensive experiments demonstrate that \textsc{FLUID} attains \(91\%\) accuracy on the GLAMI-1M benchmark, significantly outperforming prior baselines and exhibiting strong resilience to label noise, long-tail class imbalance, and semantic heterogeneity. Targeted ablation studies corroborate both the individual and synergistic benefits of the proposed components, positioning \textsc{FLUID} as a scalable, noise-resilient solution for multimodal product classification.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
