See markov chain monte carlo algorithms

kernel methods, 59, 106
Kullback–Leibler distance, 86

latent variables, 337, 343
LBG algorithm, 384
learning vector quantisation, 390
likelihood ratio, 7
linear discriminant analysis, 123–158

transformation mixture models

least mean squared error procedures, mixture sampling, 160

two-class algorithms, 124–144 multidimensional scaling by transformation,

linear discriminant function, see 352

piecewise, 21 nearest class mean classifier, 21, 36

logistic discrimination, 158–163 neural networks, 169–202

nonlinear feature extraction
multidimensional scaling by transformation, 351
nonparametric discrimination
histogram approximations, 84 Bayesian networks, 88–91

marginal independence, 84

histogram method, 82, 119
variable cell, 83
kernel methods, 106–116, 119

primary monotone condition, 349
principal components analysis, see feature extraction, principal components

93 discrimination, normal-based

normal distribution, 454 models, quadratic discriminant

EM algorithm, 42, 78 random variables

for RBF initialisation, 184 autocorrelation, 451

function, 35 ratio of uniforms method, 62

receiver operating characteristic, 15,

parameter estimation, 431–435 maximum likelihood, 433

regularisation, 155, 174–175
regularised discriminant analysis, 37, 78 reject option, 6, 9, 13
rejection sampling, 62
reliability, see performance assessment,

perceptron, see linear discriminant analysis,

perceptron criterion sampling