Continuous Sigmoidal Belief Networks Trained using Slice Sampling

By Brendan J. Frey. 1996

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Nonlinear Statistical Learning with Truncated Gaussian Graphical Models (2016, methodology): “To overcome the linearity of GGMs, Frey (1997), Hinton & Ghahramani (1997), and Frey & Hinton (1999) proposed to apply nonlinear transformations to Gaussian hidden variables.”
Online and Linear-Time Attention by Enforcing Monotonic Alignments (2017, background): “(12), as was done e.g. in (Frey, 1997; Salakhutdinov & Hinton, 2009; Foerster et al., 2016).”
The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables (2016, background): “(Frey, 1997) introduced nonlinear Gaussian units which computed their activation by passing Gaussian noise with the mean and variance specified by the input to the unit through a nonlinearity, such as the logistic function.”
Learning the Structure of Deep Sparse Graphical Models (2009, methodology): “To achieve a wide range of possible behaviors, we use the nonlinear Gaussian belief network (NLGBN) (Frey, 1997; Frey and Hinton, 1999) framework.”
Research on exchange rate forecasting based on deep belief network (2017, unknown): “This stochastic behavior is similar to the noisy unit introduced in [17].”
Forecasting exchange rate with deep belief networks (2011, background): “Such behavior is similar to the noisy unit in [17].”
Variational Learning in Nonlinear Gaussian Belief Networks (1999, methodology): “In (Frey 1997a) and (Frey 1997b), it was shown that Markov chain Monte Carlo and a variational method could be used to train small networks of these units.”
Generative models for discovering sparse distributed representations. (1997, methodology): “Because the network can select which basis functions are appropriate for the data, it can 5By using a few linear regimes we can crudely approximate units whose output is a smooth non-linear function of y (Frey, 1997) and still perform exact Gibbs sampling.”
Slice Sampling (2000, methodology): “Some such applications will involve tailor-made sampling schemes for particular models — for instance, Frey (1997) used slice sampling successfully to sample for latent variables in a neural network.”