Colloquium
Rank-Based Support Vector Machines for Highly Imbalanced Data Using Nominated Samples
Mohammad Jafari Jozani, University of Manitoba, Winnipeg, Canada
21.06.2023
4:15 pm - 5:45 pm
LMU Munich, Department of Statistics and via zoom
The talk proposes a novel approach, MaxNS, that tackles highly imbalanced binary classification using expert opinions and rank information. Biasing training samples towards the minority class, it employs rank-based Hinge and Logistic loss functions.