Stochastic Gradient Langevin Dynamics
GPTKB entity
Statements (23)
Predicate | Object |
---|---|
gptkbp:instanceOf |
mathematical optimization
|
gptkbp:abbreviation |
gptkb:SGLD
|
gptkbp:basedOn |
gptkb:Stochastic_gradient_descent
gptkb:Langevin_dynamics |
gptkbp:category |
Markov chain
|
gptkbp:feature |
scalable to large datasets
adds Gaussian noise to gradients combines optimization and sampling |
gptkbp:field |
gptkb:artificial_intelligence
deep learning statistics |
https://www.w3.org/2000/01/rdf-schema#label |
Stochastic Gradient Langevin Dynamics
|
gptkbp:proposedBy |
gptkb:Yee_Whye_Teh
gptkb:Max_Welling |
gptkbp:publicationYear |
2011
|
gptkbp:purpose |
sampling from posterior distribution
|
gptkbp:relatedTo |
gptkb:Metropolis-adjusted_Langevin_algorithm
gptkb:Stochastic_Gradient_Hamiltonian_Monte_Carlo gptkb:Stochastic_gradient_descent |
gptkbp:usedIn |
gptkb:machine_learning
Bayesian inference |
gptkbp:bfsParent |
gptkb:Max_Welling
|
gptkbp:bfsLayer |
6
|