Daniel@0: Daniel@0: Daniel@0: Daniel@0: Netlab Reference Manual demard Daniel@0: Daniel@0: Daniel@0: Daniel@0:

demard Daniel@0:

Daniel@0:

Daniel@0: Purpose Daniel@0:

Daniel@0: Automatic relevance determination using the MLP. Daniel@0: Daniel@0:

Daniel@0: Synopsis Daniel@0:

Daniel@0:

Daniel@0: demmlp1

Daniel@0: Daniel@0: Daniel@0:

Daniel@0: Description Daniel@0:

Daniel@0: This script demonstrates the technique of automatic relevance Daniel@0: determination (ARD) using a synthetic problem having three input Daniel@0: variables: x1 is sampled uniformly from the range (0,1) and has Daniel@0: a low level of added Gaussian noise, x2 is a copy of x1 Daniel@0: with a higher level of added noise, and x3 is sampled randomly Daniel@0: from a Gaussian distribution. The single target variable is determined Daniel@0: by sin(2*pi*x1) with additive Gaussian noise. Thus x1 is Daniel@0: very relevant for determining the target value, x2 is of some Daniel@0: relevance, while x3 is irrelevant. The prior over weights is Daniel@0: given by the ARD Gaussian prior with a separate hyper-parameter for Daniel@0: the group of weights associated with each input. A multi-layer Daniel@0: perceptron is trained on this data, with re-estimation of the Daniel@0: hyper-parameters using evidence. The final values for the Daniel@0: hyper-parameters reflect the relative importance of the three inputs. Daniel@0: Daniel@0:

Daniel@0: See Also Daniel@0:

Daniel@0: demmlp1, demev1, mlp, evidence

Daniel@0: Pages: Daniel@0: Index Daniel@0:

Daniel@0: