Decentralized Conformal Novelty Detection via Quantized Model Exchange

TL;DR

This paper introduces a decentralized novelty detection framework that uses quantized model exchange to control false discovery rate across heterogeneous data sources, ensuring privacy and reducing communication costs.

Contribution

It proposes a novel method for sharing low-precision surrogate models for non-conformity scores, with theoretical guarantees for FDR control in a decentralized setting.

Findings

Empirical results confirm theoretical FDR control guarantees.

The approach maintains competitive statistical power.

Communication cost is significantly reduced.

Abstract

This work studies decentralized novelty detection with global false discovery rate (FDR) control across heterogeneous composite null distributions, without sharing the raw data due to privacy and bandwidth considerations. We propose a framework based on the exchange of quantized surrogate models, allowing independent agents to share low-precision representations of locally learned non-conformity score functions. We prove that evaluating data against these quantized composite scores preserves conditional exchangeability, providing rigorous finite-sample guarantees for global FDR control. Empirical studies on synthetic datasets confirm our theoretical results, demonstrating that the proposed approach maintains competitive statistical power while drastically reducing the communication cost.