SLANG: Fast Structured Covariance Approximations for Bayesian Deep Learning with Natural Gradient.
Aaron MishkinFrederik KunstnerDidrik NielsenMark SchmidtMohammad Emtiyaz KhanPublished in: NeurIPS (2018)
Keyphrases
- deep learning
- natural gradient
- independent component analysis
- blind source separation
- unsupervised learning
- learning rate
- machine learning
- bayesian networks
- maximum likelihood
- mental models
- faster convergence
- signal processing
- data sets
- policy gradient
- pairwise
- feature extraction
- weakly supervised
- approximation methods
- data mining