Threshold optimization for weighted voting classifiers

Weighted voting classifiers considered in this paper consist of N units each providing individual classification decisions. The entire system output is based on tallying the weighted votes for each decision and choosing the one which has total support weight exceeding a certain threshold. Each individual unit may abstain from voting. The entire system may also abstain from voting if no decision support weight exceeds the threshold. Existing methods of evaluating the reliability of weighted voting systems can be applied to limited special cases of these systems and impose some restrictions on their parameters. In this paper a universal generating function method is suggested which allows the reliability of weighted voting classifiers to be exactly evaluated without imposing constraints on unit weights. Based on this method, the classifier reliability is determined as a function of a threshold factor, and a procedure is suggested for finding the threshold which minimizes the cost of damage caused by classifier failures (misclassification and abstention may have different price.) Dynamic and static threshold voting rules are considered and compared. A method of analyzing the influence of units' availability on the entire classifier reliability is suggested, and illustrative examples are presented. © 2003 Wiley Periodicals, Inc. Naval Research Logistics 50: 322–344, 2003.

[1]  J. von Neumann,et al.  Probabilistic Logic and the Synthesis of Reliable Organisms from Unreliable Components , 1956 .

[2]  Algirdas Avizienis,et al.  The STAR (Self-Testing And Repairing) Computer: An Investigation of the Theory and Practice of Fault-Tolerant Computer Design , 1971, IEEE Transactions on Computers.

[3]  William H. Press,et al.  Numerical recipes in C. The art of scientific computing , 1987 .

[4]  Akhil Kumar,et al.  Voting mechanisms in distributed systems , 1991 .

[5]  Rong-Jaye Chen,et al.  An algorithm for computing the reliability of weighted-k-out-of-n systems , 1994 .

[6]  S. M. Wu,et al.  Case Studies on Modeling Manufacturing Processes Using Artificial Neural Networks , 1995 .

[7]  Hussein Almuallim,et al.  Simplification of majority-voting classifiers using binary decision diagrams , 1996, Systems and Computers in Japan.

[8]  D. Elmakis,et al.  Redundancy optimization for series-parallel multi-state systems , 1998 .

[9]  Vijay V. Raghavan,et al.  Feature selection and effective classifiers , 1998, KDD 1998.

[10]  Anibal T. de Almeida,et al.  Combining a fuzzy classifier with classifiers based on statistic moments , 1998 .

[11]  Hoang Pham,et al.  Weighted voting systems , 1999 .

[12]  Hoang Pham.  Reliability analysis for dynamic configurations of systems with three failure modes , 1999 .

[13]  Gregory Levitin,et al.  Joint redundancy and maintenance optimization for multistate series–parallel systems , 1999 .

[14]  Gregory Levitin,et al.  Importance and sensitivity analysis of multi-state systems using the universal generating function method , 1999 .

[15]  Gregory Levitin,et al.  Multistate series-parallel system expansion-scheduling subject to availability constraints , 2000, IEEE Trans. Reliab..

[16]  Gregory Levitin,et al.  Reliability optimization for weighted voting system , 2001, Reliab. Eng. Syst. Saf..