Prediction of the Recognition Reliability using Clustering Results

The recognition reliability problem when realization of the recognition experiments is connected to some difficulties (some operative conditions or material expenditures) is considered. For estimation of recognition reliability clustering process is used. Particularly, clustering results must unambiguously determine the number of clusters and their contents. The main requirement is validity of clustering results, which depends on the cardinal number of set of realizations in learning sample. The set of realizations must be representative as far as possible. In case of satisfaction of the above-mentioned conditions it is possible using of other procedures differing from the presented.