Informative Bayesian Neural Network Priors for Weak Signals

Mathematics

• Domain Knowledge 100%
• Bayesian Networks 79%
• Neural Networks 59%
• Hyperparameters 53%
• Sparsity 44%
• Proportion 38%
• Local Features 32%
• Gaussian Mixture 29%
• Scale Mixture 27%
• Signal-to-noise Ratio 26%
• Cross-validation 23%
• Scale Parameter 21%
• Encoding 21%
• Prior distribution 21%
• Relevance 20%
• High-dimensional 17%
• Prediction 17%
• Knowledge 16%
• Gradient 16%
• Optimization 16%
• Model 6%

Engineering & Materials Science

• Neural networks 38%
• Genetics 28%
• Signal to noise ratio 17%
• Tuning 16%