Cost-sensitive Fuzzy Multiple Kernel Learning for imbalanced problem

Multiple Kernel Learning With Minority Oversampling for Classifying Imbalanced Data

“…Set the gradients of the Lagrangian function (19) for variables f m , b, ξ − , ξ + and ξ * to zero, we obtain the following: (20) in the Lagrangian function (19), the dual problem of MKLMO can be obtained as follows:…”

“…The second category of the CIL method is the algorithmperspective group, which mainly reduces the sensitivity of an algorithm to the class imbalance by modifying the misclassification costs of different instances through cost-sensitive learning [19]. For example, in the diagnosis of coronavirus disease 2019 (COVID- 19), it is much more expensive to misdiagnose COVID-19 as influenza than the opposite misdiagnosis as COVID-19 may threaten the life of a patient and spreads extremely fast. In this case, it is difficult to make an optimal decision by minimizing the total probability of error.…”