58 commits

Author	SHA1	Message	Date
Alan Saul	20227fb2ac	Made more numerically stable in a hope that it will work and I will find a bug...	2013-05-30 16:17:37 +01:00
Alan Saul	23ed2a2d15	Lots of name changing and went through all likelihood gradients again	2013-05-29 17:33:06 +01:00
Alan Saul	117c377d13	Ripped out all things Laplace parameter estimation, starting again with new tactic	2013-05-29 14:02:03 +01:00
Alan Saul	cad2c271eb	merged with devel	2013-05-14 16:26:57 +01:00
Alan Saul	5472c5c6ba	Almost have likelihood gradients working but kernels still way off	2013-05-13 18:36:02 +01:00
James Hensman	529d7534ca	fixed the bug in stick	2013-05-13 10:26:54 +01:00
Alan Saul	9500b12b53	Working on putting callback to update laplace in callback	2013-05-08 20:53:23 +01:00
Alan Saul	6c4866662c	Seem to have gradients much closer now	2013-05-08 16:05:01 +01:00
Alan Saul	2bb794120d	Merged with upstream	2013-05-07 13:41:05 +01:00
Alan Saul	a52c20f470	Added a debug examples	2013-05-07 13:35:47 +01:00
Alan Saul	f95666a8f9	Merging	2013-05-06 10:15:39 +01:00
James Hensman	0ddf308d11	fixed has_uncertain_inputs weirdness	2013-05-01 11:11:23 +01:00
James Hensman	83510e7515	first attemot at the new constraint framework	2013-04-30 11:25:57 +01:00
Alan Saul	bd346643db	Merged with master	2013-04-29 18:17:43 +01:00
James Hensman	52ba8e4ba3	remo0ved slices from models ... slices are now handles by special indexing kern parts, such as coregionalisation, independent_outputs. The old slicing functionality has been removed simply to clean up the code a little. Now that input_slices still exist (and will continue to be useful) in kern.py. They do need a little work though, for the psi-statistics	2013-04-28 17:22:04 +01:00
Neil Lawrence	ac842d51e6	cmu_mocap() example mostly working except some fiddling with axes for visualization. Also changes to naming of scaling and offset parameters in GP.py and deal with the case where the scale parameter is zero.	2013-04-27 10:39:55 +01:00
Alan Saul	267a8e427c	Adding gradients, shapes starting to make sense	2013-04-19 17:41:01 +01:00
Alan Saul	1420aa532c	Attempted to introduce gradient methods, won't work yet I doubt	2013-04-19 12:23:00 +01:00
Max Zwiessele	10703e4774	demo changed, bgplvm still broken	2013-04-18 17:59:01 +01:00
James Hensman	45403105f9	changed X_uncertainty for X_variance (in the code) for consistency with actual naming (in the printing)	2013-04-10 09:34:15 +01:00
James Hensman	128b2b90ef	more fixing of the predictive variance (correct for full_cov now)	2013-03-13 10:37:46 +00:00
James Hensman	eb82d6a255	corrected the predictive variance for Gaussian likelihoods	2013-03-13 09:58:39 +00:00
James Hensman	6a330db253	americanized spellings	2013-03-11 13:26:39 +00:00
Alan Saul	95cedc7e4e	fixed merge conflicts	2013-03-11 12:26:44 +00:00
Alan Saul	12d6f5056b	removed keyname partial	2013-03-11 12:15:59 +00:00
Ricardo Andrade	5dbc5bdb6e	Plotting functions for sparse_GP added	2013-03-11 11:40:17 +00:00
Nicolo Fusi	81810d3a7b	JH bugfix for slices	2013-03-06 16:00:14 +00:00
James Hensman	62424e5947	fixed levels in GP.plot	2013-03-01 11:40:18 +00:00
James Hensman	4674ed4558	added optional number of contour levels to the 2D plotting in GP.plot	2013-03-01 11:13:55 +00:00
Ricardo Andrade	6f55358a45	Bug found and fixed in plots for normalized X.	2013-02-08 14:00:13 +00:00
Nicolas	2abaafd882	Modifications made to tutorial due to changes in GPy	2013-02-07 13:04:29 +00:00
Nicolas	b0f6495ed4	Fixed bug in the product of kernels with tied parameters	2013-02-06 09:52:54 +00:00
Nicolas	a1568ca1c8	Few more fix to the plotings and predictions	2013-02-05 18:57:30 +00:00
Nicolas	642b1449e1	Small fixes to ploting	2013-02-05 14:23:51 +00:00
Nicolas	5b19d8609a	Many modifications in GP plots to make it work	2013-02-05 12:27:12 +00:00
Ricardo Andrade	049e60f16b	var[:,None] added in full_cov = false, sparse_GP	2013-02-04 12:01:27 +00:00
Ricardo Andrade	5e2baf1919	Changes in plotting functions.	2013-02-01 17:58:21 +00:00
James Hensman	f7d2fc6ca4	Merge branch 'newGP' of github.com:SheffieldML/GPy into newGP	2013-02-01 17:13:10 +00:00
James Hensman	5447d6fbfc	Assorted work on combining the EP and sparse methods	2013-02-01 17:12:45 +00:00
Ricardo Andrade	0a8686d7c0	EPEM is running.	2013-02-01 15:14:11 +00:00
Ricardo Andrade	f941d629e6	James' debugging of the EP/GP interface ... It seems that the GP-EP algorithm works now.	2013-02-01 13:45:55 +00:00
Ricardo Andrade	eb04cbed63	merged changes in likelihood_functions (James)	2013-02-01 13:32:13 +00:00
Ricardo Andrade	182c4c7d64	So many changes	2013-02-01 13:17:17 +00:00
Ricardo Andrade	de53917039	Merge branch 'newGP' of github.com:SheffieldML/GPy into newGP	2013-01-31 15:34:54 +00:00
James Hensman	3a558d8244	merged conflict	2013-01-31 15:30:58 +00:00
Ricardo Andrade	2b40ee6f7e	predictive_values implemented in EP	2013-01-31 15:30:57 +00:00
James Hensman	4211cad89d	re-indented the plot function	2013-01-31 15:04:24 +00:00
James Hensman	d077d28fd1	very basic functionality is now working	2013-01-31 15:02:34 +00:00
Ricardo Andrade	9feae765dc	predictive_mean changed to predictive_values	2013-01-31 14:43:32 +00:00
James Hensman	ea0802d938	much tidying and breakage in the GP class	2013-01-31 12:00:57 +00:00