Missing mass approximations for the partition function of stimulus driven Ising models.
Robert HaslingerDemba E. BaRalf GaluskeZiv WilliamsGordon PipaPublished in: Frontiers Comput. Neurosci. (2013)
Keyphrases
- partition function
- structured prediction
- random fields
- markov random field
- undirected graphical models
- parameter estimation
- upper bound
- closed form
- bayesian framework
- maximum margin
- belief propagation
- probability density function
- importance sampling
- approximate inference
- structure learning
- upper and lower bounds
- graphical models
- maximum entropy
- probabilistic model
- conditional random fields
- missing data
- exponential family
- maximum likelihood estimation
- efficient learning
- generative model
- least squares
- probability distribution