Core Concepts
Latent Variables
Latent variables are factors that are not directly measured but are inferred from observable data. In machine learning, they capture hidden structure—for example, the topics in a document collection (topic modeling) or the style of an image (variational autoencoders).
Understanding latent space is key to generative models: a model's latent space encodes compressed representations of data, and sampling from it generates new outputs. Manipulating latent variables enables controlled generation.
Authority Links
Related Terms
Model Components
Generative Model
AI model that learns to generate new data instances resembling the training distribution.
Techniques & Methods
Topic Modeling
Statistical method for discovering abstract topics within large document collections.
Model Components
Embeddings
Dense numerical vectors that represent text, images, or other content in a high-dimensional space where semantically similar items are geometrically close — the foundational data structure for semantic search and RAG retrieval.
Model Components
Generative Adversarial Network (GAN)
Framework training two competing networks—a generator and discriminator—to produce realistic synthetic data.