autoenc_dd.m

模式识别工具箱,希望对大家有用!

%AUTOENC_DD Auto-Encoder data description.
% 
%       W = autoenc_dd(A,fracrej,N)
% 
% Train an Auto-Encoder network with N hidden units.
% 
% See also datasets, mappings, dd_roc

% Copyright: D. Tax, R.P.W. Duin, [e-mail]
% Faculty of Applied Physics, Delft University of Technology
% P.O. Box 5046, 2600 GA Delft, The Netherlands


function [W,out] = autoenc_nn(a,fracrej,N)

if nargin < 3 | isempty(N), N = 5; end
if nargin < 2 | isempty(fracrej), fracrej = 0.05; end
if nargin < 1 | isempty(a) % empty nndd
  W = mapping(mfilename,{fracrej,N});
  return
end

if isa(fracrej,'double')           %training

  if ~isa(a,'dataset')              %train on training set
    error('I need a dataset to train');
  end
  a = target_class(a);     % make sure a is an OC dataset
  [nlab,lablist,m,k,c] = dataset(a);

  [nrx,dim] = size(+a);
  minmax = [min(+a)' max(+a)'];
  net = newff(minmax,[N dim],{'tansig','purelin'},'trainlm');
  net = init(net);
  net.trainParam.show = inf;
  net.trainParam.lr = 0.01;
  net.trainParam.goal = 1e-5;
  net = train(net,+a',+a');

  % obtain the threshold:
  aout = sim(net,+a');
  d = sum((a-aout').^2,2);
  thr = -threshold(-d,fracrej);

  %and save all useful data:
  W.net = net;
  W.scale = mean(d);
  W.threshold = thr;
  W = mapping(mfilename,W,str2mat('target','outlier'),k,c);

else                               %testing

  [W,classlist,type,k,c] = mapping(fracrej);  % unpack
  [nlab,lablist,m,k,c,p] = dataset(a);

  %compute distance:
  out = sim(W.net,+a')';
  out = [sum((a-out).^2,2) ones(m,1)*W.threshold];

  % map to probability
  newout = dist2dens(out,W.scale);

  W = dataset(newout,getlab(a),classlist,p,lablist);
end
return