Exact upper and lower bounds on the misclassification probability
arXiv:1712.00812v41 citations
Originality Synthesis-oriented
AI Analysis
This work provides theoretical bounds for classification error, which is incremental as it builds on prior entropy-based results.
The paper derived exact lower and upper bounds on the optimal misclassification probability for finite-class classification, expressed using total variation norms of class sub-distribution differences, and compared these to existing entropy-based bounds by Feder and Merhav.
Exact lower and upper bounds on the best possible misclassification probability for a finite number of classes are obtained in terms of the total variation norms of the differences between the sub-distributions over the classes. These bounds are compared with the exact bounds in terms of the conditional entropy obtained by Feder and Merhav.