Skip to content

Latest commit

 

History

History
28 lines (25 loc) · 1.27 KB

progress.md

File metadata and controls

28 lines (25 loc) · 1.27 KB

Progress

This document details some of the progress made on the library

TODO

  • Add tests from DSDGP
  • Add method to initialize model params
  • Test DSDGP on toy example (add to docs)
  • Test DSDGP and its optins further
  • Work on improving documentation
  • Look into multitask learning extensions

Design Choices

  • Sequential can be intialised with a partial list of layers, e.g. [InputLayer, HiddenLayer, HiddenLayer]. Should we disallow this in favour of the correct layer structure, i.e. OutputLayer at the end?
  • Currently Sequential only works by modifying GPflow source code. If pull request accepted nothing to be done here. Else, design needs to be changed.
  • InputLayer takes an initial value for Z, maybe better to disallow this and add it to the model object instead
  • Sequential has an _initialized private attribute. Turn this into property.
  • So far, multikernel layers have a set of inducing inputs for each kernel. One idea is to allow a shared set of inducing inputs for all kernels (this could act as a form of regularasation)

Known Bugs

  • Prediction with full covariance doesn't work because of numerical issues
  • Optimisation does not work
  • Gpflow bug: methods decorated with autoflow do not take named arguments