% this is a template input file for Gaussian Mixture Modeling %GMM parameters %number of clusters k = 1 2 3 4 5 6 %number of restarts of basic EM algorithm (to avoid local max.) restarts=10 %maximum number of iterations in a single EM restart max_iterations = 100 %convergence criterion: when relative increase in log-likelihood %drops below this value the convergence is declared and algorithm stops. precision = 1e-4 %CV parameters %number of CV runs. This many times will the algorithm split the data %into training/test partitions and do the basic GMM procedure cv_runs=10 %training size. This number is between 0.0 and 1.0 and tells the algorithm %what fraction of the data to use for training (the remainder is for testing). training_size = 0.6 %files %the main data file of n*d white-space separated values. Rows are individual %data points, columns are individual dimensions (variables). data_file : data.txt %output file output_file : res.txt % results sought. %'Summary' will print out details from the input file, %'Parameters' will print out fitted parameters (weights, means, covariance matrices) % on the training data. %'Classification' will print out probabilistic and deterministic classification % of test data, %'loglikelihood' will print out per-run log likelihood, together with mean for each % of the models. output: summary, parameters, classification, loglikelihood %include comments. If this line is omitted there will be no text in output file %(this might be useful for direct import into some other programs). comments %comments = no