wolffd@0: wolffd@0: wolffd@0: wolffd@0: Netlab Reference Manual nethess wolffd@0: wolffd@0: wolffd@0: wolffd@0:

nethess wolffd@0:

wolffd@0:

wolffd@0: Purpose wolffd@0:

wolffd@0: Evaluate network Hessian wolffd@0: wolffd@0:

wolffd@0: Synopsis wolffd@0:

wolffd@0:

wolffd@0: h = neterr(w, net, x, t)
wolffd@0: [h, varargout] = neterr(w, net, x, t, varargin)
wolffd@0:

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

wolffd@0: Description wolffd@0:

wolffd@0: wolffd@0:

h = nethess(w, net, x, t) takes a weight vector w and a network wolffd@0: data structure net, together with the matrix x of input wolffd@0: vectors and the matrix t of target vectors, and returns the wolffd@0: value of the Hessian evaluated at w. wolffd@0: wolffd@0:

[e, varargout] = nethess(w, net, x, t, varargin) also returns any additional wolffd@0: return values from the network Hessian function, and passes additional arguments wolffd@0: to that function. wolffd@0: wolffd@0:

wolffd@0: Example wolffd@0:

wolffd@0: wolffd@0:

In evidence, this function is called once to compute the wolffd@0: data contribution to the Hessian wolffd@0:

wolffd@0: 
wolffd@0: [h, dh] = nethess(w, net, x, t, dh);
wolffd@0:

wolffd@0: wolffd@0: and again to update the Hessian for new values of the hyper-parameters wolffd@0:

wolffd@0: 
wolffd@0: h = nethess(w, net, x, t, dh);
wolffd@0:

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

wolffd@0: See Also wolffd@0:

wolffd@0: neterr, netgrad, netopt

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

wolffd@0: