Numerically Stable Sparse Gaussian Processes via Minimum Separation using Cover Trees

Install develop version

> pip install -r requirements.txt
> pip install -e .

To use UCI datasets you would also need to follow the installation instructions of the bayesian_benchmarks package.

Usage

The experiments script used for generating tables and plots are prefixed with paper_*:

paper_cli_uci.py

Command line interface for prediction using SGPR and CDGP models with loaded hyperparameters.

Usage: paper_cli_uci.py [OPTIONS] COMMAND [ARGS]...

  This is a core command for all CLI functions.

Options:
  -mc, --model-class [sgpr|cdgp]  [required]
  -p, --precision DTYPE           [required]
  -j, --jitter FLOAT              [required]
  -c, --config-dir PATH
  --jit / --no-jit
  --help                          Show this message and exit.

Commands:
  covertree
  greedy
  kmeans
  kmeans2
  oips
  uniform