Next: References Up: G-LVQa combination of Previous: G-LVQ

Results

G-LVQ has been tested on easy synthetic problems (gaussian cluster classification), hard synthetic problems (Hart's problem) and real problems (macromolecule image classification and sonar image classification). It usually obtains dictionnaries that are shorter and more accurate than other evolutionary or static (like simple LVQ ) approaches.


jmerelo@
Mon Nov 21 11:09:09 MET 1994