BAYESIAN NON-LINEAR MODELING FOR THE PREDICTION COMPETITION

By David John Cameron MacKay. 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

Variational Dropout Sparsifies Deep Neural Networks (2017, methodology): “Then we describe Variational Dropout, a recently proposed Bayesian regularization technique (Kingma et al., 2015).”
Gaussian Process Kernels for Pattern Discovery and Extrapolation (2013, background): “A type of “auto-matic relevance determination” (MacKay et al., 1994; Tipping, 2004) takes place during training, minimizing the eﬀect of extraneous components in the proposed model, through the complexity penalty in the marginal likelihood (Rasmussen & Williams, 2006).”
Bayesian Optimization and Semiparametric Models with Applications to Assistive Technology (2014, methodology): “However, allowing for separate values of each performs Automatic Relevance Determination [MacKay, 1994a] as it allows the model to select the relevant dimensions of the input during learning.”
Structural Modelling with Sparse Kernels (2002, methodology): “However, in the commonly employed evidence framework a maximum a posteriori (MAP) approach is adopted (MacKay, 1994; Bishop, 1995).”
Gaussian Process Covariance Kernels for Pattern Discovery and Extrapolation (2013, background): “Gaussian processes with a squared exponential kernel are simply smoothing devices: the only covariance structure that can be learned from data is the length-scale `, which determines how quickly a Gaussian process function varies with x. Assuming Gaussian noise, one can analytically infer a…”
Information Theory, Inference, and Learning Algorithms (2004, methodology): “…but we don’t know which, we can define a new model with multiple hyperparameters that captures the idea of uncertain input variable relevance (MacKay, 1994b; Neal, 1996; MacKay, 1995b); these models then infer automatically from the data which are the relevant input variables for a…”
Comparing Families of Dynamic Causal Models (2010, background): “Finally, we comment on the broader issue of comparison of discrete models (the ‘Discrete’ approach adopted in this work) versus a hierarchical approach embodying Automatic Relevance Determination (ARD) in which irrelevant connections are ‘switched off’ during model fitting [41] (for the case of DCMs”
Bayesian model search for mixture models based on optimizing variational bounds (2002, methodology): “This kind of hyperprior for input variable selection, called the Automatic Relevance Determination (ARD), is proposed by (MacKay, 1994; Neal, 1996) and has been successfully used in several models (Bishop, 1999; Ghahramani & Beal, 2000; Tipping, 2000).”
Bayesian approach for neural networks--review and case studies (2001, methodology): “In this section, we discuss a simple hierarchical prior for the MLP weights, called automatic relevance determination (ARD) (MacKay, 1994; Neal, 1996, 1998).”
Blind source separation problem in GPS time series (2016, methodology): “To this aim, the Bayesian framework allows us to use the automatic relevance determination (ARD) method (e.g., MacKay 1994).”
A review of methods to match building energy simulation models to measured data (2014, methodology): “Bayesian calibration methods [73,74] can be used to naturally incorporate these uncertainties in the calibration process including the remaining uncertainty over the ﬁ tted parameters.”
Reinforcement learning for factored Markov decision processes (2002, methodology): “R==Random driver; Q==flat Q-learning; QS==flat Q-learning, stochastic policy; LL==localist linear; LM==localist MLP; DLA==distributed linear with ARD; DMA==distributed MLP with ARD; H==human driver . . . . . . . . . . . . . . . . . . . . .”
Variance Networks: When Expectation Does Not Meet Your Expectations (2018, methodology): “Automatic Relevance Determination prior p(wij) = N (wij | 0, λ(2)ij) [23, 20] has been previously applied to linear models with DSVI [32], and can be applied to Bayesian neural networks without changes.”
The generalized LASSO (2004, background): “For a general introduction to the ARD principle the reader is referred to [8].”
Bayesian Neural Networks for Internet Traffic Classification (2007, methodology): ““Automatic Relevancy Detection” (ARD) priors have been use d to infer weights of neural networks for prediction problem s ( [36], [37]).”