latent variable models
Latent variable models are statistical models that assume the existence of unobserved variables influencing observable data. They are used for tasks like clustering and dimensionality reduction.
- Can Diffusion Models Disentangle? A Theoretical Perspective
- Fisher meets Feynman: score-based variational inference with a product of experts
- Guarantees for Alternating Least Squares in Overparameterized Tensor Decompositions
- Learning Latent Variable Models via Jarzynski-adjusted Langevin Algorithm
- Learning Parameterized Skills from Demonstrations
- On the Value of Cross-Modal Misalignment in Multimodal Representation Learning
- Revisiting Multi-Agent World Modeling from a Diffusion-Inspired Perspective
- Time-Evolving Dynamical System for Learning Latent Representations of Mouse Visual Neural Activity