TY - JOUR
T1 - Fuzzy k-nearest neighbors with monotonicity constraints
T2 - Moving towards the robustness of monotonic noise
AU - González, Sergio
AU - García, Salvador
AU - Li, Sheng Tun
AU - John, Robert
AU - Herrera, Francisco
N1 - Funding Information:
This work was supported by the Spanish Ministry of Economy and Competitiveness under Grant TIN2017-89517-P and a research scholarship (FPU) given to Sergio González by the Spanish Ministry of Education, Culture and Sports. The third author was supported in part by MOST, Taiwan, R.O.C., under MOST 105-2410-H-006-038-MY3. We wish that this work serves as a tribute to our friend Bob for his life dedicated to academics and research, our collaboration and friendship, we miss you.
Publisher Copyright:
© 2021 Elsevier B.V.
PY - 2021/6/7
Y1 - 2021/6/7
N2 - This paper proposes a new model based on Fuzzy k-Nearest Neighbors for classification with monotonic constraints, Monotonic Fuzzy k-NN (MonFkNN). Real-life data-sets often do not comply with monotonic constraints due to class noise. MonFkNN incorporates a new calculation of fuzzy memberships, which increases robustness against monotonic noise without the need for relabeling. Our proposal has been designed to be adaptable to the different needs of the problem being tackled. In several experimental studies, we show significant improvements in accuracy while matching the best degree of monotonicity obtained by comparable methods. We also show that MonFkNN empirically achieves improved performance compared with Monotonic k-NN in the presence of large amounts of class noise.
AB - This paper proposes a new model based on Fuzzy k-Nearest Neighbors for classification with monotonic constraints, Monotonic Fuzzy k-NN (MonFkNN). Real-life data-sets often do not comply with monotonic constraints due to class noise. MonFkNN incorporates a new calculation of fuzzy memberships, which increases robustness against monotonic noise without the need for relabeling. Our proposal has been designed to be adaptable to the different needs of the problem being tackled. In several experimental studies, we show significant improvements in accuracy while matching the best degree of monotonicity obtained by comparable methods. We also show that MonFkNN empirically achieves improved performance compared with Monotonic k-NN in the presence of large amounts of class noise.
UR - http://www.scopus.com/inward/record.url?scp=85100899334&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85100899334&partnerID=8YFLogxK
U2 - 10.1016/j.neucom.2019.12.152
DO - 10.1016/j.neucom.2019.12.152
M3 - Article
AN - SCOPUS:85100899334
SN - 0925-2312
VL - 439
SP - 106
EP - 121
JO - Neurocomputing
JF - Neurocomputing
ER -