Fin-Fed-OD: Federated Outlier Detection on Financial Tabular Data

TL;DR

This paper introduces Fin-Fed-OD, a federated learning approach using autoencoders to improve outlier detection in financial data without sharing sensitive information, showing significant performance gains.

Contribution

It presents a novel federated outlier detection method leveraging representation learning with autoencoders, enhancing detection of unknown anomalies while preserving data privacy.

Findings

Improved outlier detection accuracy on financial datasets

Effective detection of unknown anomalies in a federated setting

Model parameters sharing preserves data confidentiality

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 novel method leveraging representation learning and federated learning techniques 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, preserving data privacy. The efficacy of our proposed method is evaluated on two standard financial tabular datasets and an image dataset 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.