DFD-HR：通过层级路由学习实现可泛化的深度伪造检测
DFD-HR: Generalizable Deepfake Detection via Hierarchical Routing Learning
把 PEFT 从「少调参数」推进到「调对参数」：层级的自适应早退剪枝加词元级 Spearman 排序筛选，可直接迁移到任何基于 CLIP 的伪造/异常检测或细粒度分类任务中，作为提升跨域泛化的通用微调策略。
原文依据 we propose integrating both layer-level and token-level routing to maximize representational synergy
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论随着生成模型的快速发展，构建可泛化的深度伪造检测器变得日益重要。通过参数高效微调（PEFT）仅更新少量参数来适配视觉基础模型（如 CLIP），已被证明对可泛化检测十分有效。然而，「更少参数」训练的成功引出一个重要问题：尽管只微调了少量参数，现有基于 PEFT 的检测器是否真正利用了最具信息量的参数、同时剔除了冗余参数以获得更好的泛化性？在本工作中，我们超越标准 PEFT，提出一种在层与词元两个层级上联合优化的策略。由于不同层的潜变量特征刻画了不同的语义抽象，而同一层内的词元传达了多样的伪造线索，我们提出将层级路由与词元级路由相结合，以最大化表征协同。具体而言，在层级上，我们引入「早层剪枝」，这是一种自适应截断机制，使模型能够针对不同类型的实例自适应地学习不同的前向深度。在词元级上，「词元选择」由 Spearman 排序损失引导，以过滤与伪造学习无关的词元，使模型聚焦于最具判别性的线索。此外，我们采用统一的混合专家（MoE）架构，以鼓励多样性，从而降低模型对特定伪造类型过拟合的潜在风险。大量基准测试结果验证了我们设计的有效性，并表明本方法优于现有最先进方法。
英文原文摘要 1679 字符
Developing generalizable deepfake detectors has become increasingly important with the rapid advancement of generative models. Adapting visual foundation models (VFMs), e.g., CLIP, through parameter-efficient finetuning (PEFT), with only a small subset of parameters updated, has been proven highly effective for generalizable detection. However, the success of "fewer-parameters" training raises an important question: although only a few parameters are tuned, have existing PEFT-based detectors truly exploited the most informative ones while eliminating redundant parameters for better generalization? In this work, we move beyond standard PEFT by proposing a joint optimization strategy that operates at both the layer and token levels. Since latent features across layers capture different semantic abstractions and tokens within the same layer convey varied forgery cues, we propose integrating both layer-level and token-level routing to maximize representational synergy. Specifically, at the layer level, we introduce "Early Layer Pruning", an adaptive truncation mechanism that enables the model to adaptively learn distinct forward depths for different types of instances. At the token level, "Token Selection" is guided by the Spearman rank loss to filter tokens irrelevant to forgery learning, enabling the model to focus on the most discriminative cues. Furthermore, a unified MoE architecture is applied that encourages diversity and thus reduces the potential model's overfitting to specific forgery types. Extensive benchmarking results demonstrate the effectiveness of our designs and show the superior performance of our method over existing state-of-the-arts.