# Multimodal Fusion **Entity class:** Sensor And Multimodal-Fusion Layer concept **Multimodal Fusion** is a data-integration concept for bringing heterogeneous observations into a common, provenance-aware analytic environment. Multimodal Fusion belongs in the ontology because AI can compute over this structure while analysts retain responsibility for interpretation and authorized action. ## Counterterrorism predictive-graph role The page records capability and analytic function. Any claim of a named operational deployment remains subject to the [[wiki/Counterterrorism Predictive Graph Evidence Ladder|evidence ladder]]. The analytic state should distinguish ground truth, observed evidence, inferred state, and predicted state. ## Relationships - **domain router:** [[wiki/Intelligence Fusion|Intelligence Fusion]]. - **ontology neighbors:** [[wiki/Multisensor Fusion|Multisensor Fusion]] and [[wiki/Information Fusion|Information Fusion]]. - **synthesis:** [[wiki/Counterterrorism Predictive Graph|Counterterrorism Predictive Graph]] and [[wiki/Predictive Intelligence Loop|Predictive Intelligence Loop]]. ## Sources - [IARPA — DIVA program (accessed September 23, 2026)](https://www.iarpa.gov/research-programs/diva) - [IARPA — AGILE technical overview (accessed September 23, 2026)](https://www.iarpa.gov/images/research-programs/AGILE/AGILE_ModSim_-_Final.pdf) **As of:** 2026-09-23