Extend approach to other modalities

Background: The Identifiable Variational Dynamic Factor Model (iVDFM) is currently formulated for multivariate time series data.

Question / Future Work: Extend the framework of conditionally identifiable latent representations and dynamic factor modeling to other data modalities, such as language or image sequences, to see if the identifiability techniques can be successfully adapted.

Why It Matters: Extending the model to other modalities tests the generalizability of the core innovation on identifiability coupled with dynamics beyond numerical time series.

Evidence: Further research may explore other conditional contexts (e.g., learned or task-specific embeddings), relax the dynamical or prior structure while preserving identifiability, and extend the approach to other modalities such as language or image sequences.