NICTA/revrand

Issues

• Publish Nlopt wrappers of old revrand project as separate package

  #133 opened 4 years ago by dschmitz89
  4

• How to Simulate A la Carte

  #132 opened 5 years ago by FBoyang
  2

• problems with demos

  #131 opened 7 years ago by AIAdventures
  4

• Finish report

  #104 opened 8 years ago by dsteinberg
  0

• Make ARD length-scale bases return generator gradients instead of tensors

  #130 opened 8 years ago by dsteinberg
  0

• Extensive profiling

  #106 opened 8 years ago by dsteinberg
  0

• Gaussian Spectral Mixture using RFF approximation

  #93 opened 8 years ago by dsteinberg
  0

• Implement random arc-cosine features

  #128 opened 8 years ago by dsteinberg
  1

• implement even more bases

  #87 opened 8 years ago by dsteinberg
  0

• Use scikit learn "check_estimator" in tests

  #129 opened 8 years ago by dsteinberg
  1

• Explore softplus as a replacement to log for optimising positive parameters

  #127 opened 8 years ago by dsteinberg
  1

• Check uncoverml revrand gridsearch traceback

  #124 opened 8 years ago by basaks
  2

• Make nice __repr__ methods for all of the objects in revrand

  #122 opened 8 years ago by dsteinberg
  0

• Use slices in calculating composed basis gradients rather than zero padding?

  #123 opened 8 years ago by dsteinberg
  0

• Scalar regularizer is not the same as regularizer per-basis type for kernel composition

  #120 opened 8 years ago by dsteinberg
  0

• Basis and Likelihood classes should use the parameters in the initialiser, and not require the inputs of the same parameters into the methods

  #121 opened 8 years ago by dsteinberg
  1

• Basis functions should have "gain" parameters

  #119 opened 8 years ago by dsteinberg
  1

• Investigate numerical issues in tests

  #116 opened 8 years ago by dsteinberg
  1

• Simple global optimisation for SLM (and GLM)

  #99 opened 8 years ago by dsteinberg
  0

• Import a few more revrand classes/functions into the main namespace?

  #113 opened 8 years ago by dsteinberg
  0

• Trivial item, base.py comment unclear meaning

  #118 opened 8 years ago by zunzun
  1

• Should the GLM and SLM set RandomState of input objects?

  #117 opened 8 years ago by dsteinberg
  1

• StandardLinearModel and GeneralisedLinearModel do not work with sklearn.base.clone

  #115 opened 8 years ago by dsteinberg
  0

• Add "check_random_state" functionality to the random bases

  #114 opened 8 years ago by dsteinberg
  0

• Numerical overflow in logtrick

  #110 opened 8 years ago by dsteinberg
  2

• remove/archive legacyGP?

  #112 opened 8 years ago by dsteinberg
  0

• Implement predict_proba methods for the SLM and GLM

  #105 opened 8 years ago by dsteinberg
  3

• Improve Optimisation and Model tests

  #102 opened 8 years ago by dsteinberg
  1

• Implement Auto-Encoding VB for the GLM

  #108 opened 8 years ago by dsteinberg
  0

• combination basis error with GLM

  #109 opened 8 years ago by MushroomHunting
  3

• Initialise the weights of the GLM with the MAP solution

  #107 opened 8 years ago by dsteinberg
  0

• Improve documentation

  #92 opened 8 years ago by dsteinberg
  1

• Return named tuple from glm and slm learn functions

  #97 opened 8 years ago by dsteinberg
  2

• Do we need to use append_or_extend with the optimisers?

  #103 opened 8 years ago by dsteinberg
  0

• Convert all demos to notebooks

  #101 opened 8 years ago by dsteinberg
  0

• Variance of prediction results with BLR and GLM (and a few questions)

  #98 opened 8 years ago by MushroomHunting
  3

• We are still getting Non PSD errors in solve_posdef

  #100 opened 8 years ago by dsteinberg
  1

• Implement predict_likelihood in glm

  #94 opened 8 years ago by dsteinberg
  0

• Sampling in GLM predict functions has slowed down

  #96 opened 8 years ago by dsteinberg
  0

• Implement three parameter beta likelihood

  #95 opened 8 years ago by dsteinberg
  0

• Write up GLM and SLM in the report

  #88 opened 8 years ago by dsteinberg
  1

• Allow per-datum likelihood arguments in GLM

  #86 opened 8 years ago by dsteinberg
  0

• Cannot install :(

  #90 opened 8 years ago by ltiao
  7

• Reconsider logging and printing

  #80 opened 8 years ago by dsteinberg
  1

• Do we still need safe divide (safediv) with safechol?

  #85 opened 8 years ago by dsteinberg
  1

• Replace jitchol with safechol

  #84 opened 8 years ago by dsteinberg
  0

• Make the output of predict_meanvar etc more transparent

  #82 opened 9 years ago by dsteinberg
  0

• __call__ members are not documented in sphinx

  #79 opened 9 years ago by dsteinberg
  0

• Instead of inputting initial basis paramaters in bparams lists, input them as part of basis function construction

  #83 opened 9 years ago by dsteinberg
  2

• Replicate scipy/numpy heirarchy

  #81 opened 9 years ago by dsteinberg
  0