Principled Pruning of Bayesian Neural Networks through Variational Free Energy Minimization.
Jim BeckersBart van ErpZiyue ZhaoKirill KondrashovBert de VriesPublished in: CoRR (2022)
Keyphrases
- energy minimization
- neural network
- image segmentation
- energy function
- graph cuts
- low level vision
- global minimum
- markov random field
- belief propagation
- problems in computer vision
- free energy
- dual decomposition
- global minimization
- active contour model
- curve evolution
- bayesian networks
- genetic algorithm
- max flow min cut
- search space
- max flow
- multiscale
- weighted constraint satisfaction
- posterior probability
- image dependent
- level set
- graphical models
- optical flow
- min cut max flow
- machine learning
- min cut
- early vision
- global minima