Notebooks

Basic

• Bayesian Gaussian process latent variable model (Bayesian GPLVM)
• Basic (binary) GP classification model
• Monitoring Optimisation
• Basic (Gaussian likelihood) GP regression model

Advanced

• More details on models with many observation points
• Change points
• Convolutional Gaussian Processes
• A simple demonstration of coregionalization
• Faster predictions by caching
• Stochastic Variational Inference for scalability with SVGP
• Heteroskedastic Likelihood and Multi-Latent GP
• Manipulating kernels
• MCMC (Markov Chain Monte Carlo)
• Multiclass classification
• Multi-output Gaussian processes in GPflow
• Natural gradients
• Optimizers
• Ordinal regression
• Variational Fourier Features in the GPflow framework
• Gaussian process regression with varying output noise

Tailor

• Custom mean functions: metalearning with GPs
• Mixing TensorFlow models with GPflow
• Kernel Design
• Likelihood Design
• Mixture Density Networks in GPflow
• Models with observed and latent variables
• Updating model with new data

Theory

• Comparing FITC approximation to VFE approximation
• Derivation of SGPR equations
• Sanity checking when model behaviours should overlap
• Discussion of the GP marginal likelihood upper bound
• Derivation of VGP equations

Understanding

• Architecture
• Manipulating GPflow models
• Utilities