Nip Rumors in the Bud: Retrieval-Guided Topic-Level Adaptation for Test-Time Fake News Video Detection

TL;DR

This paper introduces RADAR, a novel test-time adaptation framework for fake news video detection that leverages retrieval-guided methods and stable references to adapt to unseen topics and emerging events.

Contribution

RADAR is the first framework to enable test-time adaptation for fake news videos using retrieval-guided topic-level adaptation and stable reference alignment.

Findings

RADAR outperforms existing methods in detecting fake news videos on unseen topics.

It effectively adapts to fast-changing label distributions during test time.

RADAR demonstrates strong on-the-fly adaptation capabilities in real-world scenarios.

Abstract

Fake News Video Detection (FNVD) is critical for social stability. Existing methods typically assume consistent news topic distribution between training and test phases, failing to detect fake news videos tied to emerging events and unseen topics. To bridge this gap, we introduce RADAR, the first framework that enables test-time adaptation to unseen news videos. RADAR pioneers a new retrieval-guided adaptation paradigm that leverages stable (source-close) videos from the target domain to guide robust adaptation of semantically related but unstable instances. Specifically, we propose an Entropy Selection-Based Retrieval mechanism that provides videos with stable (low-entropy), relevant references for adaptation. We also introduce a Stable Anchor-Guided Alignment module that explicitly aligns unstable instances' representations to the source domain via distribution-level matching with their stable references, mitigating severe domain discrepancies. Finally, our novel Target-Domain Aware Self-Training paradigm can generate informative pseudo-labels augmented by stable references, capturing varying and imbalanced category distributions in the target domain and enabling RADAR to adapt to the fast-changing label distributions. Extensive experiments demonstrate that RADAR achieves superior performance for test-time FNVD, enabling strong on-the-fly adaptation to unseen fake news video topics.