Approximate inference systems (AxIS): end-to-end approximations for energy-efficient inference at the edge.
Soumendu Kumar GhoshArnab RahaVijay RaghunathanPublished in: ISLPED (2020)
Keyphrases
- efficient inference
- end to end
- approximate inference
- exact inference
- probabilistic inference
- graphical models
- variational methods
- factor graphs
- conditional random fields
- gaussian process
- parameter estimation
- belief propagation
- bayesian networks
- message passing
- latent variables
- human pose estimation
- markov random field
- hidden variables
- dynamic bayesian networks
- fully connected
- image processing
- hidden markov models
- structured prediction
- high quality
- higher order
- graph structure