Recent developments of the minimum entropy algorithm
Satosi Watanabé, Tsuguchika Kaminuma · 2003
This work has two purposes: it is a tutorial survey of methods based on the heuristic principle of entropy minimization; and it introduces a novel algorithm IMPAL, that makes possible to decide the class affiliation of a new arrival when multiclass paradigms are given in advance. Two examples are given to demonstrate the relevance of IMPAL. >