vfind

This program will repeatedly run the initialization, training and testing cycle for Self-Organizing Map algorithm. In the following the training file name, the test file name (that can be the same) and the map save file name are asked. After them the type of map topology is asked, as well as the type of neighborhood function. The x- and y-dimension of the map should be integers and prefereably x-dimension should be larger than y-dimension. The training is done in two parts. First an ordering phase that is usually shorter than the following converging phase. The number of training cycles, the training rates and the radius of the adaptation area are asked separately for both phases. The fixed point qualifiers and weighting qualifiers are used if the corresponding parameters were given. The quantization error is computed for each map and the best map (smallest quantization error) is saved to the given file. If the verbose parameter allows the quantization error is given for each separate trial. After the answers have been given the training begins and depending on the size of problem it may take a long time. Give the number of trials: