Generative models · MLD-0916
Posterior collapse
A failure in latent-variable models where the learned latent representation carries little information about the observed input.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Detect an expressive decoder ignoring its latent variables.
Original caveat
Low reconstruction loss alone may not reveal collapse.
Context and aliases
Level: Advanced. Stages: Architecture · Training · Inference.
Tags: DL · Generative AI · Vision.
Aliases: None recorded.
Related concepts
Original references
- Denoising Diffusion Probabilistic Models ↗ · Consulted
- Deep Learning: online textbook ↗ · Consulted
- Flow Matching for Generative Modeling ↗ · Consulted
- Auto-Encoding Variational Bayes ↗ · Consulted
- Original reference URL ↗