GP software Version 0.133 Sunday 06 Sep 2009 at 06:07 The GP toolbox is an implementation of the GPs that uses the Pseudo-Input method of Snelson and Ghahramani (NIPS 2005) for sparsification as well as extensions given by Quinonero-Candela and Rasmussen (JMLR 2005). Version 0.133 ------------- Updates for running a GPLVM/GP using the data's inner product matrix for Interspeech synthesis demos. Version 0.132 ------------- Remove bug in gpExtractParam for sparse models where scale parameters were in the wrong place. Moved in some examples from the oxford toolbox. Added Titsias's variational sparse approximation. Version 0.131 ------------- Changes to allow compatibility with SGPLVM and NCCA toolboxes. Version 0.13 ------------ Changes to allow more flexibility in optimisation of beta. Version 0.12 ------------ Various minor changes for enabling back constraints in hierarchical GP-LVM models. Version 0.11 ------------ Changes include the use of the optimiDefaultConstraint('positive') to obtain the function to constrain beta to be positive (which now returns 'exp' rather than 'negLogLogit' which was previously the default). Similarly default optimiser is now given by a command in optimiDefaultOptimiser. Version 0.1 ----------- The first version which is spun out of the FGPLVM toolbox. The corresponding FGPLVM toolbox is 0.15. MATLAB Files ------------ Matlab files associated with the toolbox are: gpPosteriorGradMeanVar.m: Gadient of the mean and variances of the posterior at points given by X. gpCovGradsTest.m: Test the gradients of the likelihood wrt the covariance. gpSubspaceOut.m: gpUpdateKernels.m: Update the kernels that are needed. gpPointLogLikelihood.m: Log-likelihood of a test point for a GP. demGpSample.m: Simple demonstration of sampling from a covariance function. gpPosteriorMeanCovarTest.m: Test the gradients of the mean and covariance. demSpgp1d2.m: Do a simple 1-D regression after Snelson & Ghahramani's example. gpToolboxes.m: Load in the relevant toolboxes for GP. demGpCovFuncSample.m: Sample from some different covariance functions. gpObjective.m: Wrapper function for GP objective. demSpgp1d5.m: Do a simple 1-D regression after Snelson & Ghahramani's example. demSilhouetteGp1.m: Model silhouette data with independent RBF GPs. gpComputeM.m: Compute the matrix m given the model. gpComputeAlpha.m: Update the vector `alpha' for computing posterior mean quickly. gpReadFromFile.m: Load a file produced by the C++ implementation. demInterpolation.m: Demonstrate Gaussian processes for interpolation. gpPosteriorVar.m: Variances of the posterior at points given by X. gpScaleBiasGradient.m: Compute the log likelihood gradient wrt the scales. demSilhouettePlot.m: gpSubspaceExtractParam.m: demSpgp1dPlot.m: Plot results from 1-D sparse GP. demOptimiseGp.m: Shows that there is an optimum for the covariance function length scale. gpSubspaceCreate.m: demOptimiseGpTutorial.m: Shows that there is an optimum for the covariance function length scale. gpPosteriorSample.m: Create a plot of samples from a posterior covariance. demSilhouettePlotTrue.m: Plot the true poses for the silhouette data. demStickGp1.m: Demonstrate Gaussian processes for regression on stick man data. gpUpdateAD.m: Update the representations of A and D associated with the model. gpObjectiveGradient.m: Wrapper function for GP objective and gradient. demSpgp1d4.m: Do a simple 1-D regression after Snelson & Ghahramani's example. gpGradient.m: Gradient wrapper for a GP model. gpExpandParam.m: Expand a parameter vector into a GP model. gpWriteToFile.m: Write a file to be read by the C++ implementation. gpOut.m: Evaluate the output of an Gaussian process model. demSilhouetteAverage.m: Shows the average of the poses. demSpgp1d1.m: Do a simple 1-D regression after Snelson & Ghahramani's example. demRegression.m: Demonstrate Gaussian processes for regression. gpLogLikelihood.m: Compute the log likelihood of a GP. gpSample.m: Create a plot of samples from a GP. demSilhouetteGp2.m: Model silhouette data with independent MLP GPs. gpDataIndices.m: Return indices of present data. gpLogLikeGradients.m: Compute the gradients for the parameters and X. gpCovGrads.m: Sparse objective function gradients wrt Covariance functions for inducing variables. demSpgp1d3.m: Do a simple 1-D regression after Snelson & Ghahramani's example. gpDeconstruct.m: break GP in pieces for saving. gpCreate.m: Create a GP model with inducing varibles/pseudo-inputs. gpSubspaceExpandParam.m: gpExtractParam.m: Extract a parameter vector from a GP model. gpSubspaceOptimise.m: gpOptions.m: Return default options for GP model. gpReconstruct.m: Reconstruct an GP form component parts. gpDisplay.m: Display a Gaussian process model. gpTest.m: Test the gradients of the gpCovGrads function and the gp models. demGpCov2D.m: Simple demonstration of sampling from a covariance function. gpPosteriorGradMeanCovar.m: Gadient of the mean and variances of the posterior at points given by X. demSilhouetteLinear1.m: Model silhouette data with independent linear models. gpPosteriorMeanCovar.m: Mean and covariances of the posterior at points given by X. gpBlockIndices.m: Return indices of given block. gpReadFromFID.m: Load from a FID produced by the C++ implementation. gpPosteriorMeanVar.m: Mean and variances of the posterior at points given by X. gpMeanFunctionGradient.m: Compute the log likelihood gradient wrt the scales. gpOptimise.m: Optimise the inducing variable based kernel.