Bayesian methods for adaptive models

By David John Cameron MacKay. 1992

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

Deep Gaussian Processes (2012, methodology): “nalized allowing a variational lower bound on the likelihood to be computed. The appropriate size of the latent space can be computed using automatic relevance determination (ARD) priors [Neal, 1996, MacKay et al., 1992]. Damianou et al. [2011] extended this approach by placing a Gaussian process pr”
Uncertainty-guided Continual Learning with Bayesian Neural Networks (2019, methodology): “ns the information learned on previous tasks using an almost-binary mask per previous tasks. Bayesian approaches: Using Bayesian approach in learning neural networks has been studied for few decades (MacKay, 1992b;a). Several approaches have been proposed for Bayesian neural networks, based on, e.g.”
Predictive Uncertainty Estimation via Prior Networks (2018, methodology): “ep Learning through the deﬁnition of baselines, tasks and metrics [13] and the development of practical methods for estimating uncertainty. One class of approaches stems from Bayesian Neural Networks [14, 15]. Traditionally, these approaches have been computationally more demanding and conceptually ”
Three Factors Influencing Minima in SGD (2017, methodology): “7 The Laplace approximation is a common approach used to approximate integrals, used for example by MacKay (1992) and Kass & Raftery (1995).”
Support Vector Data Description (2004, methodology): “In classification or regression problems a more advanced Bayesian approach can be used for detecting outliers (Bishop, 1995; MacKay, 1992; Roberts & Penny, 1996).”
Predicting motor vehicle crashes using Support Vector Machine models. (2008, methodology): “To circumvent this problem, Bayesian neural networks (BNN) have been proposed by a few researchers (Liang, 2005; Mackay, 1992; Neal, 1995).”
Probable networks and plausible predictions - a review of practical Bayesian methods for supervised neural networks (1995, methodology): “As shown in (Gull 1989;MacKay 1992a), the maximum evidence = MP satis es the following implicitequation: 1= MP =Xi wMPi 2= (22) 19where wMP is the parameter vector which minimizes the objective function M = ED + EW , and is the `number of well-determined parameters', given by = k Trace(A 1):…”
Application of support vector machine and relevance vector machine to determine evaporative losses in reservoirs (2012, methodology): “…to maximize the term p ( y/˛, 2 ) : p ( y / ˛, 2 ) D s p (y/w, 2) p w/˛ dw D ( 2 N 2 /√∣∣ 2 C A 1T∣∣ ) ð exp { 1 2 yT ( 2 C A 1T) 1 y} 12 Maximisation of this quantity is known as the type II maximum likelihood method (Berger, 1985; Wahba, 1985) or the ‘evidence for hyperparameter’ (Mackay, 1992).”
A Bayesian approach to identification of hybrid systems (2004, methodology): “This approach to model comparison is similar to the Bayesian framework which was used in [13] for neural networks.”
Quantifying model form uncertainty in Reynolds-averaged turbulence models with Bayesian deep neural networks (2018, methodology): “techniques for obtaining Bayesian statistics include the placement of distributions over network weights and sampling with Monte Carlo methods to approximate statistical bounds [32, 33] as well as ensemble methods [34, 35].”
Unsupervised Quality Estimation for Neural Machine Translation (2020, methodology): “Uncertainty quantification in deep learning typically relies on the Bayesian formalism (MacKay, 1992; Graves, 2011; Welling and Teh, 2011; Gal and Ghahramani, 2016; Tran et al., 2019).”