wolffd@0: wolffd@0: wolffd@0: wolffd@0: Netlab Reference Manual demev1 wolffd@0: wolffd@0: wolffd@0: wolffd@0:

demev1 wolffd@0:

wolffd@0:

wolffd@0: Purpose wolffd@0:

wolffd@0: Demonstrate Bayesian regression for the MLP. wolffd@0: wolffd@0:

wolffd@0: Synopsis wolffd@0:

wolffd@0:

wolffd@0: demev1

wolffd@0: wolffd@0: wolffd@0:

wolffd@0: Description wolffd@0:

wolffd@0: The problem consists an input variable x which sampled from a wolffd@0: Gaussian distribution, and a target variable t generated by wolffd@0: computing sin(2*pi*x) and adding Gaussian noise. A 2-layer wolffd@0: network with linear outputs is trained by minimizing a sum-of-squares wolffd@0: error function with isotropic Gaussian regularizer, using the scaled wolffd@0: conjugate gradient optimizer. The hyperparameters alpha and wolffd@0: beta are re-estimated using the function evidence. A graph wolffd@0: is plotted of the original function, the training data, the trained wolffd@0: network function, and the error bars. wolffd@0: wolffd@0:

wolffd@0: See Also wolffd@0:

wolffd@0: evidence, mlp, scg, demard, demmlp1

wolffd@0: Pages: wolffd@0: Index wolffd@0:

wolffd@0: