Sparsifying Bayesian neural networks with latent binary variables and normalizing flows.
Lars Skaaret-LundGeir StorvikAliaksandr HubinPublished in: CoRR (2023)
Keyphrases
- binary variables
- neural network
- random variables
- knapsack problem
- latent variables
- continuous variables
- bayesian networks
- piecewise linear
- linear program
- limited memory
- problems in computer vision
- mixed integer programming
- energy function
- posterior distribution
- linear constraints
- np hard
- maximum likelihood
- linear programming relaxation
- mixed integer
- valid inequalities
- real time
- bayesian inference
- convex functions
- parameter estimation
- linear programming
- graphical models
- probability distribution
- computational complexity