KI-Glossar

📖

Begriffe

Adversarial Regularization

Method using a discriminator to force the latent distribution to match a target distribution, avoiding the posterior collapse problem.

📖

Begriffe

Latent Variable Modeling

Probabilistic approach assuming that observed data are generated from unobserved latent variables following a specific distribution.

📖

Begriffe

Encoding Distribution

Parametric distribution learned by the encoder that approximates the true posterior of latent variables conditioned on input data.

📖

Begriffe

Decoding Distribution

Conditional distribution modeled by the decoder that generates output data from sampled latent variables.