A random decision classifier based on mini-max principle to improve the performance of classifiers that have poor efficiency. This algorithm is based on the concept given in Digital Signal Processing for Wireless Communication by Dr. E.S.Gopi
The code requires two input files one "datatx.txt" is the actual label of a sample while "datarx.txt" contains the classifier's output for given samples. find_init.m compares the actual labels with the classifier's output while find_perf.m compares actual labels with the final output of the random classifier.
Run the file mini_max.m. The output of the random classifier is stored in "datadetected.txt"
#Application The application of this technique includes binary classification problems where there are many unknown parameters which results in significant drop in accuracy.