839 commits

Author	SHA1	Message	Date
Max Zwiessele	c6d9ebf08d	[licensing] replaced licensing with BSD, and erfcx	2015-09-07 17:26:27 +01:00
Max Zwiessele	929cf0a489	[more coverage] and predictive var fixes	2015-09-07 16:52:59 +01:00
Max Zwiessele	63e1fa05e4	[mean func] added in GPRegression	2015-09-07 14:10:36 +01:00
Alan Saul	a156cc5dc9	Allowed gpyified var gauss	2015-09-07 14:00:04 +01:00
Max Zwiessele	87ee4ae548	[MRD] fixed mrd for new structure	2015-09-04 16:43:26 +01:00
Max Zwiessele	8fa5b67089	[spgp minibatch] linear calls the right psicomps and add kernel	2015-09-04 13:30:15 +01:00
Max Zwiessele	79dd821424	[mrd] fixes for updates on psi2	2015-09-04 11:07:39 +01:00
Max Zwiessele	9ddec5bc70	[merge] for spgp minibatch and psi NxMxM	2015-09-04 10:37:29 +01:00
Max Zwiessele	cb024f0bf8	[spgp minibatch] added new routine for psi NxMxM, much faster, little bigger mem footbprint	2015-09-04 10:33:32 +01:00
Alan Saul	b73932c350	Small edits for linear kernel	2015-09-03 13:32:59 +03:00
Max Zwiessele	839e3dc6f0	[magnification] plot_magnification expanded	2015-09-03 09:33:07 +01:00
Max Zwiessele	ca60ad3195	[magnification] mostly plotting and some model corrections for _predictive_variable	2015-09-02 15:46:40 +01:00
Max Zwiessele	74a7c6398b	[merge]	2015-09-02 14:34:23 +01:00
Ricardo	4bec78704b	Note added to the docs	2015-09-02 13:32:39 +01:00
Ricardo	7a5872e11d	Model uses the new HeteroscedasticGaussian likelihood	2015-09-02 12:15:06 +01:00
Max Zwiessele	70a9a26d7e	[core] updating system, security branching	2015-09-02 09:06:17 +01:00
Alan Saul	81ef734908	Reindented, did some profiling which looks promising	2015-09-02 10:46:30 +03:00
Alan Saul	50b9e4dc82	corrected caching for psi derivatives	2015-09-01 19:17:56 +03:00
Alan Saul	3818aa3745	Try calculating dL_dpsi1*psi1 individually for each dimension as we go along	2015-08-31 14:13:35 +03:00
Alan Saul	c83f56723e	passing psi statistics	2015-08-30 22:22:27 +03:00
Alan Saul	1a2b194aa2	Removed comment	2015-08-28 14:23:44 +03:00
Alan Saul	b89196c7cf	Checking kerngrads full_value	2015-08-28 14:03:14 +03:00
Alan Saul	e12575c0c6	Gradients for X seem to match	2015-08-28 11:47:10 +03:00
Alan Saul	afba8d5c5e	Made sparse gp work again	2015-08-28 10:07:19 +03:00
Alan Saul	5f119d0918	Tidied but broken	2015-08-27 19:08:11 +03:00
Alan Saul	4143f00540	Optimizing missing data model, needs tidying but now much faster	2015-08-27 17:56:08 +03:00
James Hensman	7ba26eda1c	the Opper-Archambeau method is now implemented as an inference method in the GPy style	2015-08-18 11:39:27 +01:00
James Hensman	ea3bfbb597	minor bugfixes in plotting: quantiles are now computed using predict_kw ... correctly	2015-08-18 08:00:47 +01:00
Alan Saul	65b6e54d5c	Merge branch 'devel' of github.com:SheffieldML/GPy into devel	2015-08-17 16:06:47 +01:00
Alan Saul	070137504a	Added full cov prediction	2015-08-17 16:06:43 +01:00
James Hensman	e3fd7885bf	ENH: fixed up BCGPLVM to work with new framework	2015-08-17 11:06:45 +01:00
beckdaniel	20a1e8d3c7	some cleaning on WarpedGP code	2015-08-05 19:18:25 +01:00
beckdaniel	995de0f399	added a new test which tries to replicate Snelson's toy 1D but NR seems to diverge...	2015-08-05 19:05:48 +01:00
beckdaniel	76bc0bec25	added initial test for warped gps using identity function	2015-08-05 10:50:59 +01:00
beckdaniel	2426580d4d	implemented variance using gauss-hermite	2015-08-04 18:46:10 +01:00
beckdaniel	3aa563d5ba	first try in implementing warped mean	2015-08-04 14:22:46 +01:00
beckdaniel	00a209f4cb	added predict_quantiles method to warped gps	2015-07-30 17:33:42 +01:00
beckdaniel	45321f536b	added keywords to predict in warped gps because they are used in the plotting method	2015-07-30 17:19:16 +01:00
beckdaniel	7a7d563576	commented out has_uncertain_inputs in warped gps since it breaks plotting	2015-07-30 17:09:31 +01:00
mzwiessele	d6defa6645	Merge branch 'devel' of https://github.com/SheffieldML/GPy into devel	2015-07-29 10:48:27 +02:00
Alan Saul	42a415e299	Fixed model test	2015-07-22 18:37:12 +01:00
James Hensman	efa65c864e	generalized the variatinoal Gaussian approximatino revisited code for any likelihood	2015-07-16 15:36:17 +01:00
mzwiessele	4ca4916cc0	Merge branch 'devel' of https://github.com/SheffieldML/GPy into devel	2015-06-29 10:19:43 +02:00
mzwiessele	f46b23d16f	[inference] minibatch inference needed polishing	2015-06-29 10:19:36 +02:00
Zhenwen Dai	5cc17e8754	some correction for ibp ssgplvm	2015-06-28 21:33:57 +01:00
Mike Croucher	e7d1277e62	Python 3 fixes	2015-06-23 01:26:52 -07:00
javiergonzalezh	e58cc51c47	allowing set initial noise variance for GPRegression	2015-06-15 18:42:29 +01:00
Zhenwen Dai	b5cac9af8c	[ssmrd] implement with IBP prior	2015-05-22 14:29:53 +01:00
Zhenwen Dai	afa0621488	[ssgplvm] implement IBP prior	2015-05-21 14:18:32 +01:00
Zhenwen Dai	b8508cc20c	[ssgplvm] group spike	2015-05-21 11:33:37 +01:00