Bayesian Restoration of Audio Degraded by Low-Frequency Pulses Modeled via Gaussian Process.
Hugo Tremonte de CarvalhoFlávio Rainho ÁvilaLuiz Wagner Pereira BiscainhoPublished in: CoRR (2020)
Keyphrases
- closed form
- low frequency
- gaussian process
- hyperparameters
- posterior distribution
- gaussian processes
- high frequency
- marginal likelihood
- bayesian methods
- frequency domain
- fully bayesian
- gaussian process regression
- wavelet transform
- degraded images
- expectation propagation
- image restoration
- subband
- discrete wavelet transform
- wavelet coefficients
- approximate inference
- regression model
- gaussian process models
- signal processing
- high frequency components
- image processing
- audio signal
- bayesian framework
- model selection
- bayesian networks
- bayesian inference
- latent variables
- sample size
- bit rate
- semi supervised
- active learning
- high resolution
- training set
- video sequences