Bayesian approach for neural networks--review and case studies

doi:10.1016/S0893-6080(00)00098-8

Paper

DOI: 10.1016/S0893-6080(00)00098-8

Bayesian approach for neural networks--review and case studies

Published Apr 1, 2001 · J. Lampinen, Aki Vehtari

Neural networks : the official journal of the International Neural Network Society

Q1 SJR score

351

Citations

Influential Citations

Full text Semantic Scholar

Abstract

Abstract removed due to Elsevier request; this does not indicate any issues with the research. Click the full text link above to read the abstract and view the original source.

Highly Cited

Study Snapshot

Key takeawayThe Bayesian approach for neural network learning allows for more generalization and less restrictive priors, making it more effective in regression, classification, and inverse problems.

PopulationOlder adults (50-71 years)

Sample size24

MethodsObservational

OutcomesBody Mass Index projections

ResultsSocial networks mitigate obesity in older groups.

Sign up to use Study Snapshot

Consensus is limited without an account. Create an account or sign in to get more searches and use the Study Snapshot.

Create an account

Full text analysis coming soon...

References

DOI: 10.1088/0957-0233/12/2/702

Bayesian Theory

Bayesian Theory is an excellent introduction and guide to Bayesian coherent reasoning, providing a clear understanding of the foundations and key concepts of Bayesian Statistics for metrologists.

2001·125citations·J. Bernardo et al.·Measurement Science and Technology

Measurement Science and Technology

DOI: 10.1109/IJCNN.2000.857855

On MCMC sampling in Bayesian MLP neural networks

This paper proposes a new method for choosing starting values in Bayesian MLP neural networks based on early stopping, and demonstrates the benefits of using several independent chains for MCMC sampling.

2000·25citations·Aki Vehtari et al.·Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium

Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium

DOI: 10.1162/089976600300015664

Sequential Monte Carlo Methods to Train Neural Network Models

The hybrid gradient descent/sampling importance resampling algorithm (HySIR) improves computational time and accuracy in training neural networks for on-line, nonlinear, and nongaussian signal processing applications.

2000·222citations·J. F. G. D. Freitas et al.·Neural Computation

Neural Computation

DOI: 10.1109/IJCNN.1999.832623

Bayesian neural networks with correlating residuals

Bayesian neural networks with correlating residuals can provide a more realistic model for multivariate regression problems.

1999·7citations·Aki Vehtari et al.·IJCNN'99. International Joint Conference on Neural Networks. Proceedings (Cat. No.99CH36339)

IJCNN'99. International Joint Conference on Neural Networks. Proceedings (Cat. No.99CH36339)

DOI: 10.1016/S0893-6080(99)00040-4

Bayesian neural networks for classification: how useful is the evidence framework?

The Bayesian evidence framework for neural networks can achieve similar classification accuracies to alternative methods with minimal human intervention, but requires large training examples and irrelevant variables.

1999·112citations·W. Penny et al.·Neural networks : the official journal of the International Neural Network Society

Neural networks : the official journal of the International Neural Network Society

DOI: 10.1023/A:1008820505350

Convergence assessment techniques for Markov chain Monte Carlo

This paper reviews convergence assessment techniques for Markov chain Monte Carlo methods, highlighting the mathematics underpinning these methods, and motivates future research in this area.

1998·454citations·S. Brooks et al.·Statistics and Computing

Statistics and Computing

DOI: 10.1117/12.325800

Using Bayesian neural networks to classify forest scenes

Bayesian learning methods for neural networks can effectively classify forest scenes into trees and backgrounds, reducing overfitting and improving performance compared to baseline classifiers.

1998·17citations·Aki Vehtari et al.

DOI: 10.1162/089976698300017197

Approximate Statistical Tests for Comparing Supervised Classification Learning Algorithms

The 5 2 cv test is recommended for comparing supervised classification learning algorithms due to its acceptable type I error and slightly more power than McNemar's test for once-executed algorithms.

1998·3608citations·Thomas G. Dietterich·Neural Computation

Neural Computation

DOI: 10.1007/978-1-4471-1599-1_47

Empirical Evaluation of Bayesian Sampling for Neural Classifiers

Bayesian sampling improves the generalisation performance of neural network predictors, with a critical evaluation of the automatic relevance determination (ARD) scheme for detecting irrelevant inputs.

1998·4citations·D. Husmeier et al.

Citations

DOI:

Optimizing MCMC-Driven Bayesian Neural Networks for High-Precision Medical Image Classification in Small Sample Sizes

MCMC-based Bayesian neural networks improve robustness and accuracy in medical image classification with small samples, reaching 85% for lung X-ray and 88% for breast tissue slice datasets.

2024·0citations·Mingyu Sun

DOI: 10.48550/arXiv.2409.07942

Taylor-Sensus Network: Embracing Noise to Enlighten Uncertainty for Scientific Data

The Taylor-Sensus Network (TSNet) effectively models complex, heteroscedastic noise in scientific data, improving uncertainty estimation and demonstrating superior performance in experiments.

2024·0citations·Guangxuan Song et al.·ArXiv

ArXiv

DOI: 10.1088/1748-0221/19/08/c08003

Bayesian Neural Network Variational Autoencoder Inverse Mapper (BNN-VAIM) and its application in Compton Form Factors extraction

The Bayesian Neural Network Variational Autoencoder Inverse Mapper (BNN-VAIM) effectively extracts Compton Form Factors from unpolarized Deeply virtual exclusive scattering (DVCS) cross sections, reducing uncertainty in the inverse problem.

2024·0citations·MD Fayaz Bin Hossen et al.·Journal of Instrumentation

Journal of Instrumentation

DOI: 10.3390/jmse12081275

Fatigue Load Modeling of Floating Wind Turbines Based on Vine Copula Theory and Machine Learning

The RF machine learning model accurately predicts flapwise fatigue loads at the blade root of floating wind turbines with up to 99.97% accuracy, improving their reliability and efficiency.

2024·0citations·Xinyu Yuan et al.·Journal of Marine Science and Engineering

Journal of Marine Science and Engineering

DOI: 10.48550/arXiv.2407.19287

Bayesian meta learning for trustworthy uncertainty quantification

Trust-Bayes is a novel Bayesian meta learning framework that optimizes for trustworthy uncertainty quantification in Bayesian regression, reducing sample complexity and improving accuracy.

2024·0citations·Zhenyuan Yuan et al.·ArXiv

ArXiv

DOI:

posteriordb: Testing, Benchmarking and Developing Bayesian Inference Algorithms

posteriordb is a database of 120 representative models and data sets for evaluating and comparing Bayesian inference algorithms across various target densities.

2024·3citations·Maans Magnusson et al.

DOI: 10.48550/arXiv.2407.04617

Randomized Physics-Informed Neural Networks for Bayesian Data Assimilation

The rPINN method provides faster and more accurate uncertainty quantification in inverse PDE problems compared to the HMC method, making it a promising alternative for non-linear problems.

2024·0citations·Yifei Zong et al.·ArXiv

ArXiv

DOI: 10.48550/arXiv.2406.07767

Conformalized Teleoperation: Confidently Mapping Human Inputs to High-Dimensional Robot Actions

Conformalized assistive teleoperation effectively maps low-dimensional human inputs to high-dimensional robot actions, improving robot performance in various tasks.

2024·2citations·Michelle Zhao et al.·ArXiv

ArXiv