A two-layer temporal generative model of natural video exhibits complex-cell-like pooling of simple cell outputs.
Jarmo HurriAapo HyvärinenPublished in: Neurocomputing (2003)
Keyphrases
- generative model
- probabilistic model
- spatial and temporal
- bayesian framework
- space time
- em algorithm
- dynamic textures
- semi supervised
- temporal information
- prior knowledge
- video representation
- spatio temporal
- video sequences
- video database
- posterior probability
- discriminative learning
- discriminative models
- topic models
- latent dirichlet allocation
- video streams
- machine learning
- co occurrence
- image features
- high level
- image segmentation
- image processing