Hyper-Representations as Generative Models: Sampling Unseen Neural Network Weights.
Konstantin SchürholtBoris KnyazevXavier Giró-i-NietoDamian BorthPublished in: CoRR (2022)
Keyphrases
- generative model
- neural network
- feature representations
- probabilistic model
- discriminative learning
- mixture model
- visual patterns
- artificial neural networks
- hierarchical hidden markov models
- back propagation
- linear combination
- hierarchical models
- discriminative models
- deep belief networks
- em algorithm
- posterior probability
- prior knowledge
- maximum entropy principle
- object categories
- generative and discriminative models
- expectation maximization
- semi supervised
- conditional random fields
- language modeling
- radial basis function
- representational power
- topic models
- training set
- information retrieval