Shedding the Facades, Connecting the Domains: Detecting Shifting Multimodal Hate Video with Test-Time Adaptation

2026-01-28

AI Summary

This work addresses the significant performance degradation of conventional hate video detection models caused by severe semantic drift between training and inference data due to the continuous evolution of hateful content. To mitigate this issue, the authors propose SCANNER, a novel framework that introduces test-time adaptation (TTA) to hate video detection for the first time. SCANNER identifies stable core attributes—such as gender and race—and leverages them to construct a centroid-guided, sample-level adaptive alignment mechanism. Additionally, it incorporates an intra-cluster diversity regularization to enhance model generalization. Built upon multimodal semantic modeling, the proposed method consistently outperforms existing approaches across multiple benchmarks, achieving an average improvement of 4.69% in Macro-F1 score.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Hate Video Detection (HVD) is crucial for online ecosystems. Existing methods assume identical distributions between training (source) and inference (target) data. However, hateful content often evolves into irregular and ambiguous forms to evade censorship, resulting in substantial semantic drift and rendering previously trained models ineffective. Test-Time Adaptation (TTA) offers a solution by adapting models during inference to narrow the cross-domain gap, while conventional TTA methods target mild distribution shifts and struggle with the severe semantic drift in HVD. To tackle these challenges, we propose SCANNER, the first TTA framework tailored for HVD. Motivated by the insight that, despite the evolving nature of hateful manifestations, their underlying cores remain largely invariant (i.e., targeting is still based on characteristics like gender, race, etc), we leverage these stable cores as a bridge to connect the source and target domains. Specifically, SCANNER initially reveals the stable cores from the ambiguous layout in evolving hateful content via a principled centroid-guided alignment mechanism. To alleviate the impact of outlier-like samples that are weakly correlated with centroids during the alignment process, SCANNER enhances the prior by incorporating a sample-level adaptive centroid alignment strategy, promoting more stable adaptation. Furthermore, to mitigate semantic collapse from overly uniform outputs within clusters, SCANNER introduces an intra-cluster diversity regularization that encourages the cluster-wise semantic richness. Experiments show that SCANNER outperforms all baselines, with an average gain of 4.69% in Macro-F1 over the best.

Problem

• Hate Video Detection
• Semantic Drift
• Test-Time Adaptation
• Distribution Shift
• Multimodal Hate

Innovation

• Test-Time Adaptation
• Hate Video Detection
• Semantic Invariance
• Centroid-Guided Alignment
• Intra-Cluster Diversity