Title :
Myopic Policies in Sequential Classification
Author :
Ben-Bassat, Moshe
Author_Institution :
Center for the Critically Ill, University of Southern California School of Medicine
Abstract :
Several rules for feature selection in myopic policy are examined for solving the sequential finite classification problem with conditionally independent binary features. The main finding is that no rule is consistently superior to the others. Likewise no specific strategy for the alternating of rules seems to be significantly more efficient.
Keywords :
Classification; divergence measures; feature selection; information measures; myopic policies; probability of misclassification; sequential decisions; simulation; Automata; Costs; Dynamic programming; Gold; Inference algorithms; Large-scale systems; Medical simulation; Sequential analysis; Testing; Turing machines; Classification; divergence measures; feature selection; information measures; myopic policies; probability of misclassification; sequential decisions; simulation;
Journal_Title :
Computers, IEEE Transactions on
DOI :
10.1109/TC.1978.1675054
