On the Bounds on Optimal Bayes Error in the Task of Multiple Data Sources
The paper considers the problem of pattern recognition when we have multiple data sources. We assume that for each data source there are estimated parameters of statistical distributions. The model of classification is primarily based on the Bayes rule and secondarily on the notion of interval-valued fuzzy sets. The set of possible class-conditional probability density functions is represented by an interval-valued fuzzy set. We consider the case where the uncertainty concerns the mean of Gaussian pdf. In the paper the bound on the optimal Bayess error is presented for a full probabilistic information.
Unable to display preview. Download preview PDF.