# 二、高斯过程介绍

(1)

(2)

（3）

(4)

xs =(-5:0.2:5)’; ns = size(xs,1); keps = 1e-9;

m =inline(’0.25*x.^2’);

K =inline(’exp(-0.5*(repmat(p’’,size(q))-repmat(q,size(p’’))).^2)’);

fs = m(xs) +chol(K(xs,xs)+keps*eye(ns))’*randn(ns,1);

plot(xs,fs,’.’)

# 三、后验高斯过程

(5)

（6）

（7）

（8）

# 四、训练一个高斯过程

（）                               (9)

（10）

（11）

# 六、参考文献

[1] Williams, C.K.I.: Prediction with Gaussian processes:From linear regression to linear prediction and beyond. In Jordan, M.I., ed.:Learning in Graphical Models. Kluwer Academic (1998) 599–621

[2] MacKay, D.J.C.: Gaussian processes — a replacement forsupervised neural networks? Tutorial lecture notes for NIPS 1997 (1997)

[3] Williams, C.K.I., Barber, D.: Bayesian classificationwith Gaussian processes. IEEE Transactions on PatternAnalysis and Machine Intelligence 20(12) (1998) 1342–1351

[4] Csat´o, L., Opper, M.: Sparse on-line Gaussian processes.Neural Computation 14 (2002) 641–668

[5] Neal, R.M.: Regression and classification using Gaussianprocess priors (with discussion). In Bernardo, J.M., et al., eds.: Bayesianstatistics 6. Oxford University Press (1998) 475–501*