Abstract
The application of median-related measures to fuzzy computation, especially in defuzzification and in combining fuzzy membership degrees, is presented. For near-symmetric distributions, averaging and medoid computations generally give almost identical results, and computational reasons then motivate using the former. The extra computation was required in asymmetric problems, where the choice of the medoid should be preferred. In addition, there are some distributions where the mean was undefined, but the median was defined, though such cases may not arise in practical problems.
Original language | English (US) |
---|---|
Pages (from-to) | 429-433 |
Number of pages | 5 |
Journal | Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS |
State | Published - 2000 |
Event | 19th International Confernce of the North American Fuzzy Information Processing Society-NAFIPS (PEACH FUZZ 2000) - Atlanta, GA, USA Duration: Jul 13 2000 → Jul 15 2000 |
ASJC Scopus subject areas
- General Computer Science
- General Mathematics