jamesrobertlloyd

Inference fail loop

1

If any code fails (e.g. ill conditioned covariance matrix) the experiment can get stuck in an infinite loop I observed this in the prediction code once - quick fix was...

Miscellaneous notes

2

Hessian delta of 1e-6 will not always be appropriate e.g. location parameters for unscaled data

Revisit parameter defaults and random restarts

2

Make them shift and scale with the data appropriately

Either with priors + MAP (i.e. regularisation) or a more Bayesian treatment (Laplace, sampling...?)

See readme for details - sh -> cmd - string sanitization

Fix the optimisation

1

This means both - ensuring local extrema are reached so Laplace approx valid - jittering previously learned parameter values to avoid having to multiply by constant kernels

Move unused results files so that they can be nuked if necessary

Which licence? GPL, CRAPL, MIT?

Merge LOCAL and HOME - and put fully within cblparallel

Kernel over kernels + active learning / Bayesian optimisation Priors on expressions ...