Semv framework improves multimedia verification with self evolving memory
Self-Evolving Multimedia Verification through Memory Consolidation of Contestation Experiences
MultimediaArtificial IntelligenceMultiagent Systems
Summary
Verifying whether multimedia content like images or videos is real or fake is hard because you need clear evidence, the ability to fix mistakes, and a way to remember past experiences safely. The authors created SEMV, a system where multiple parts work together to argue about the evidence, learn from human corrections, and safely update its knowledge without mixing up useful and harmful information. Their system was tested on different benchmarks and did better than existing methods, fixing most errors and using less computing power. SEMV keeps track of how it decides things, making it easier to check and improve in the future.
What this means in practice
- •For content moderation teams: Improve detection of manipulated multimedia by evolving verification processes that incorporate human feedback and prevent harmful knowledge reuse.
- •For digital forensics analysts: Use verified memory consolidation to reliably trace and update evidence and reasoning when investigating multimedia authenticity.
Authors
Truong Thanh Hung Nguyen, Vo Thanh Khang Nguyen, Hoang-Loc Cao, Phuc Ho, Truong Thinh Nguyen, Van Pham, Hung Cao
Abstract
Multimedia verification requires not only accurate decisions but also traceable evidence, reliable human correction, and safe reuse of prior experience. Existing systems often lack explicit mechanisms for revising intermediate reasoning or preventing harmful knowledge transfer. We present SEMV (Self-Evolving Multimedia Verification), a self-evolving multi-agent framework that treats provenance-bearing arguments as the interface between evidence, reasoning, human contestation, and memory. SEMV combines arena-based quantitative bipolar argumentation (A-QBAF), causal and scoped revision, and verification-gated memory consolidation with explicit conflict retention. On COSMOS benchmark, SEMV achieves 91.88% accuracy versus 89.10% for the strongest comparable baseline. Verified memory reduces negative transfer from 5.7% to 0.2%. On CTR benchmark, constructed from reviewer contestations, scoped causal revision corrects 96.7% of initial errors while saving 52.8% compute. MV2026 Grand Challenge dataset further supports evidence-grounded, temporally consistent reporting. These results show that SEMV can evolve through verified experience while keeping accumulated knowledge and subsequent decisions traceable, revisable, and contestable.