GPCsign (version 0.1.1)

update: Update of a Gaussian Process Classification (GPC) model

Description

Update a gpcm object when one or many new observations are added.

Usage

# S3 method for gpcm
update(object, newf, newXf, newX.alreadyExist,
 newnoise.var, covandmean.reestim=TRUE, multistart = 1, seed = NULL,
  lower = NULL, upper = NULL, nsimu = 100, normalize = TRUE, ...)

Value

Updated gpcm object.

Arguments

object

an object of gpcm class.

newf

a vector corresponding to the new binary observations (+/-1) at newXf locations. These locations can be new locations or existing ones.

newXf

a matrix with object@d columns representing the locations to be updated. These locations can be new locations or existing ones.

newX.alreadyExist

Boolean: indicate whether the locations newXf are all news or not. Default: TRUE, corresponding to existing locations in newX.

newnoise.var

optional scalar, nugget effect at new observations.

covandmean.reestim

should the mean and covariance parameters be re-estimated? Default is TRUE.

multistart

an optional integer indicating the number of initial points from which running the BFGS for covariance parameter optimization. Default is 1.

seed

to fix the seed, default is NULL.

lower

(see below).

upper

lower, upper: bounds for the covariance parameters (scalars or vectors), if NULL they are set to 0.2 and 3, respectively.

nsimu

an integer indicating the number of samples of the latent GP at observation points Xf to generate. Must be a non-null integer. Default is 100.

normalize

a logical parameter indicating whether to normalize the input matrix Xf. If TRUE, the matrix will be normalized using X.mean and X.std values if given; otherwise, the mean and standard deviation are calculated and used for normalization.

...

no other argument for this method

Author

Morgane MENZ, Céline HELBERT, Victor PICHENY, François BACHOC. Contributors: Naoual SERRAJI.

References

Bachoc, F., Helbert, C. & Picheny, V. Gaussian process optimization with failures: classification and convergence proof. J Glob Optim 78, 483–506 (2020). tools:::Rd_expr_doi("10.1007/s10898-020-00920-0").

Kotecha, J. H., Djuric, P. M. (1999). Gibbs Sampling Approach For Generation of Truncated Multivariate Gaussian Random Variables. IEEE Computer Society, 1757–1760.

Wilhelm, S. tmvtnorm: Truncated Multivariate Normal and Student t Distribution. R package version 1.6. https://CRAN.R-project.org/package=tmvtnorm.

Roustant, O., Ginsbourger, D. & Deville, Y. Contributors: Chevalier, C. , Richet, Y. DiceKriging: Kriging Methods for Computer Experiments. R package version 1.6.0. https://CRAN.R-project.org/package=DiceKriging.

Byrd, R. H., Lu, P., Nocedal, J. and Zhu, C. (1995). A limited memory algorithm for bound constrained optimization. SIAM Journal on Scientific Computing, 16, 1190–1208. tools:::Rd_expr_doi("10.1137/0916069").

See Also

gpcm

Examples

# ----------------------------------
# A 1D example - sinusoidal function
# ----------------------------------

# Test function
sinusoidal_function <- function(x) {
  sin(4 * pi * x)}

# Desing of Experiment Xf and the corresponding sign f
Xf <- as.matrix(c(0.07, 0.19, 0.42, 0.56, 0.81, 0.90))
f <- rep(1,length(Xf)); f[(sinusoidal_function(Xf)<0)]<- -1

# Builidng a GPC model
GPCmodel1 <- gpcm(f = f, Xf = Xf, coef.m=0, coef.cov=0.26)
print(GPCmodel1)

# New points added to the gpcm object.
newXf <- as.matrix(c(0.1,0.5,0.7, 0.95))
newf <- rep(1,length(newXf)); newf[(sinusoidal_function(newXf)<0)]<- -1

# Updating GPC model
NewGPCmodel <- update(object = GPCmodel1, newf = newf, newXf = newXf)
print(NewGPCmodel)