Skew Probabilistic Neural Networks for Learning from Imbalanced Data.
Pattern Recognition., 2025
Imbalanced datasets often lead to poor performance of standard classifiers, particularly for minority classes. This work proposes Skew Probabilistic Neural Networks (SkewPNN), a probabilistic neural network with a skew-normal kernel to better model asymmetric class distributions and improve classification accuracy. Hyperparameters are optimized using the Bat algorithm, enhancing model performance across diverse settings.
Overview of the SkewPNN Architecture.