English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Neural networks (NNs) are primarily developed within the frequentiststatistical framework. Nevertheless, frequentist NNs lack the capability toprovide uncertainties in the predictions, and hence their robustness can not beadequately assessed. Conversely, the Bayesian neural networks (BNNs) naturallyoffer predictive uncertainty by applying Bayes' theorem. However, theircomputational requirements pose significant challenges. Moreover, bothfrequentist NNs and BNNs suffer from overfitting issues when dealing with noisyand sparse data, which render their predictions unwieldy away from theavailable data space. To address both these problems simultaneously, weleverage insights from a hierarchical setting in which the parameter priors areconditional on hyperparameters to construct a BNN by applying a semi-analyticalframework known as nonlinear sparse Bayesian learning (NSBL). We call ournetwork sparse Bayesian neural network (SBNN) which aims to address thepractical and computational issues associated with BNNs. Simultaneously,imposing a sparsity-inducing prior encourages the automatic pruning ofredundant parameters based on the automatic relevance determination (ARD)concept. This process involves removing redundant parameters by optimallyselecting the precision of the parameters prior probability density functions(pdfs), resulting in a tractable treatment for overfitting. To demonstrate thebenefits of the SBNN algorithm, the study presents an illustrative regressionproblem and compares the results of a BNN using standard Bayesian inference,hierarchical Bayesian inference, and a BNN equipped with the proposedalgorithm. Subsequently, we demonstrate the importance of considering the fullparameter posterior by comparing the results with those obtained using theLaplace approximation with and without NSBL.

Sparse Bayesian neural networks for regression: Tackling overfitting and computational challenges in uncertainty quantification