Search papers, labs, and topics across Lattice.
This paper introduces DeCoFlow, a novel approach to continual anomaly detection in industrial settings that leverages Normalizing Flows (NFs) while addressing the challenge of catastrophic forgetting. By employing a structural decomposition of subnets into a frozen universal base and task-specific low-rank adapters, the method ensures parameter isolation without compromising the invertibility and Jacobian validity of NFs. The results demonstrate that DeCoFlow achieves state-of-the-art performance on benchmark datasets, with an image-level AUROC of 98.40% on MVTec-AD and zero forgetting across tasks, showcasing its effectiveness in dynamic environments.
DeCoFlow achieves a remarkable 98.40% AUROC in continual anomaly detection while ensuring zero parameter forgetting, revolutionizing how we handle sequential data in industrial applications.
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.
