prediction_snn.m

神经网络的工具箱, 神经网络的工具箱,

function [ylp, yup, y_av] = prediction_snn(c_point_pred, noise_net, nets, alpha, P)
%PREDICTION_SNN Estimate prediction intervals.
%
%  Syntax
%
%   [ylp, yup, y_av] = prediction_snn(c_pred, noise_net, nets, alpha, P)
%
%  Description
%
%   PREDICTION_SNN takes
%     c_pred    - c_prediction
%     noise_net - a net_struct predicting noise
%     nets      - [1 x M]  net_structs of trained networks with cost 
%                 function WCF_SNN.
%     alpha     - [1 x M] matrix of network weighting factors.
%     P         - [N0 x MU] matrix with inputs
%   and returns
%     ylp       - lower bound for prediction interval
%     yup       - upper bound for prediction interval
%     y_av      - weighted average output
%
%   (N0 = #inputs; MU = #patterns; M = #networks in ensemble)
%
%  Algorithm
%
%    See: T. Heskes; Practical confidence and prediction Intervals.
%    Advances in Neural Information Processing Systems 9, pages
%    176-182, Cambridge, 1997, MIT Press.
%
%  See also
%
%   C_PREDICTION_SNN, BALANCE_SNN
%

MU = size(P,2);
errf = nets(1).costFcn.fn;

[y_av, conf_err_estimate] = simff_avr_snn(nets, alpha, P);
pred_err_estimate = simff_snn(noise_net, P);

e_max = repmat(c_point_pred,1,MU) .* max(conf_err_estimate, pred_err_estimate);

[ylp, yup] = inverse_error_snn(y_av, e_max, errf);