Year: 2025
Type: conference-paper
Abstract: Anomaly detection in real-world scenarios poses challenges due to dynamic and often unknown anomaly distributions, requiring robust methods that operate under an open-world assumption. This challenge is exacerbated in practical settings, where models are employed by private organizations, precluding data sharing due to privacy and competitive concerns. Despite potential benefits, the sharing of anomaly information across organizations is restricted. This paper addresses the question of enhancing outlier detection within individual organizations without compromising data confidentiality. We propose a method called Fin-Fed-OD, a novel framework for distributed outlier detection that leverages the shared structure in representations across clients with federated learning to improve the detection of unknown anomalies. Specifically, our approach utilizes latent representations obtained from client-owned autoencoders to refine the decision boundary of inliers. Notably, only model parameters are shared between organizations, without exchanging raw data. The efficacy of our proposed method is evaluated on five standard financial tabular datasets for anomaly detection in a distributed setting. The results demonstrate a strong improvement in the classification of unknown outliers during the inference phase for each organization's model.
Institution: German Research Centre for Artificial Intelligence
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: Google
Award: G2391215542
OpenAlex ID: W7124945409