FRAUDSkill: Structured Frozen-Weight Skill Optimization for Audio Anti-Fraud Detection

Researchers introduced FRAUDSkill, a method for audio anti-fraud detection that uses structured frozen-weight skill optimization, according to the paper's headline. The approach targets large audio-language models, which the authors note can process speech and reason over fraud-related evidence but require predictions to follow a predefined label space and a structured decision protocol consisting of service-scenario identification. The work addresses the gap between those models' open-ended reasoning and the constrained label and decision structure needed for deployment.

Large audio-language models have shown promise for anti-fraud detection by directly processing speech and reasoning over fraud-related evidence. Their deployment, however, requires predictions to follow a predefined label space and a structured decision protocol consisting of service-scenario identi