Undersampling Method

Undersampling is a data preprocessing technique used to address class imbalance in datasets by reducing the number of instances in the majority class, aiming to improve the performance of machine learning models, particularly for minority classes. Current research focuses on developing sophisticated undersampling algorithms that consider data morphology, such as overlap between classes and proximity to decision boundaries, to minimize information loss and bias. This approach is valuable across diverse applications, from improving fairness in AI to enhancing the accuracy of medical image reconstruction and cybersecurity threat detection, by enabling more robust and equitable model training.