Graphical frontend for Onlinde Gaussian Process classification.
demogp_class_gui
- provides a graphical user interface for Gaussian Process
classification.
The classification is for two-dimensional inputs, these inputs are provided by the user. All OGP operations can be reached using the control buttons on the left of the demo window.
The addition of data to be classified is by using the left/right button of
the mouse over the axis. One can then click on the LEARN
button to
infer the OGP parameters.
One can set (or reset) the model hyperparameters (such as kernel type or kernel parameter) but in this case the GP inference is restarted.
This program is designed for a first contact with GP and the idea of sparsity within a full probabilistic inference. Sparsity i controlled by imposing an upper limit to the number of Basis Vectors or setting the threshold value.
ogp
, ogptrain
, demogp_class
, demogp_reg_gui
Copyright (c) Lehel Csató (2001-2004)