Caching

PHOTONAI offers a specialized caching that handles partially overlapping hyperparameter configurations for nested cross-validation splits. This is particularly useful for reusing results from expensive computations.

More generally, caching is useful whenever re-computation needs more time than loading data.

It is easily enabled by adding the cache_folder parameter to the Hyperpipe.

1
2
pipe = Hyperpipe("...",
                 cache_folder="./cache")