Bidirectional Recurrent Neural Networks as Generative Models - Reconstructing Gaps in Time Series.
Mathias BerglundTapani RaikoMikko HonkalaLeo KärkkäinenAkos VetekJuha KarhunenPublished in: CoRR (2015)
Keyphrases
- generative model
- recurrent neural networks
- chaotic time series
- echo state networks
- probabilistic model
- neural network
- reservoir computing
- feed forward
- mixture model
- discriminative models
- recurrent networks
- conditional random fields
- bayesian framework
- prior knowledge
- artificial neural networks
- hidden layer
- hierarchical hidden markov models
- generative and discriminative models
- discriminative learning
- object categories
- topic models
- maximum entropy principle
- em algorithm
- semi supervised
- graphical models
- nonlinear dynamic systems
- representational power
- deep belief networks
- genetic algorithm