SheffieldML / GPy

Gaussian processes framework in python
BSD 3-Clause "New" or "Revised" License
2.05k stars 562 forks source link

GPLVM: how to predict x_star given new y_star? #638

Closed LuckyTiger closed 6 years ago

LuckyTiger commented 6 years ago

I find https://github.com/SheffieldML/GPy/issues/395 very helpful. using m_gplvm.infer_newX(Y_new) is ok. But how does it work?

LuckyTiger commented 6 years ago

How to get predicted latent variance?

mzwiessele commented 6 years ago

I think it is all well described:

https://github.com/SheffieldML/GPy/blob/devel/GPy/inference/latent_function_inference/inferenceX.py

On 18 May 2018, at 09:18, LuckyTiger notifications@github.com wrote:

How to get predicted latent variance?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

LuckyTiger commented 6 years ago

But this don't have X's variance

mzwiessele commented 6 years ago

If the model you begin with is a Bayesian GPLVM the output of inferX will be a variational posterior, which has the means and variances attached.

LuckyTiger commented 6 years ago

But how did Lawrence's first GP-LVM paper https://papers.nips.cc/paper/2540-gaussian-process-latent-variable-models-for-visualisation-of-high-dimensional-data.pdf have uncertainty plotted?

mzwiessele commented 6 years ago

Ah, try looking into the plotting function plot_latent for details on that measure.

On 19 May 2018, at 07:49, LuckyTiger notifications@github.com wrote:

But how did Lawrence's first GP-LVM paper https://papers.nips.cc/paper/2540-gaussian-process-latent-variable-models-for-visualisation-of-high-dimensional-data.pdf have uncertainty plotted?

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or mute the thread.