DeCoFlow: 用于持续异常检测的归一化流结构分解
DeCoFlow: Structural Decomposition of Normalizing Flows for Continual Anomaly Detection
AI总结：
针对工业环境中新产品类别顺序到达且无法访问旧数据的持续异常检测问题，提出DeCoFlow方法，通过将子网络分解为冻结通用基和任务特定低秩适配器来隔离参数更新，并引入任务特定对齐、辅助耦合层和尾感知损失，在MVTec-AD和VisA数据集上达到最先进性能且零遗忘。
摘要：
在工业环境中，新产品类别顺序到达，需要在无法访问旧数据的情况下进行持续异常检测。归一化流（NFs）提供精确的密度估计，但由于跨任务的参数更新会扭曲密度流形，导致灾难性遗忘。虽然参数隔离可以防止干扰，但必须保持NFs的严格可逆性和雅可比有效性。为了满足这些要求，我们利用仿射耦合层的一个固有属性，即无论子网络参数化如何，都能保持变换的有效性。基于此，我们提出DeCoFlow，将子网络分解为冻结的通用基和任务特定的低秩适配器以隔离更新。我们进一步引入任务特定对齐、辅助耦合层和尾感知损失来补偿冻结基的刚性。DeCoFlow在MVTec-AD上实现了98.40%的图像级AUROC，在VisA上实现了93.00%，同时在正确路由下保持参数级零遗忘（0.00% FM），每个任务仅需2.27M参数。
英文摘要：
In industrial environments, new product categories arrive sequentially, requiring continual anomaly detection without access to past data. Normalizing Flows (NFs) provide exact density estimation but suffer from catastrophic forgetting as parameter updates across tasks distort the density manifold. While parameter isolation can prevent interference, it must preserve the strict invertibility and Jacobian validity of NFs. To satisfy these requirements, we exploit the inherent property that affine coupling layers maintain transformation validity regardless of subnet parameterization. Based on this, we propose DeCoFlow, which decomposes subnets into a frozen universal base and task-specific low-rank adapters to isolate updates. We further introduce Task-Specific Alignment, Auxiliary Coupling Layers, and Tail-Aware Loss to compensate for frozen-base rigidity. DeCoFlow achieves state-of-the-art image-level AUROCs of 98.40% on MVTec-AD and 93.00% on VisA, while maintaining parameter-level zero forgetting (0.00% FM under correct routing) with only 2.27M parameters per task.
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。
