Changeset 6718


Ignore:
Timestamp:
09/08/11 11:29:26 (8 years ago)
Author:
gkronber
Message:

#1474 fixed bug in alglib that resulting in random models when using the early stopping training method for multi layer perceptrons.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • trunk/sources/HeuristicLab.ExtLibs/HeuristicLab.ALGLIB/3.1.0/ALGLIB-3.1.0/dataanalysis.cs

    r4977 r6718  
    1526915269                    if( state.xupdated )
    1527015270                    {
    15271                         for(i_=0; i_<=wcount-1;i_++)
    15272                         {
    15273                             network.weights[i_] = w[i_];
     15271                        minlbfgs.minlbfgsresults(state, ref w, internalrep);
     15272                        for (i_ = 0; i_ <= wcount - 1; i_++) {
     15273                          network.weights[i_] = w[i_];
    1527415274                        }
    1527515275                        e = mlpbase.mlperror(network, valxy, valsize);
Note: See TracChangeset for help on using the changeset viewer.