Imbalanced Regression

Imbalanced regression addresses the challenge of training accurate regression models when the target variable's distribution is uneven, with some values significantly under-represented. Current research focuses on developing novel loss functions and regularization techniques to mitigate the bias towards majority classes, employing methods like contrastive learning, hierarchical classification adjustments, and probabilistic smoothing within deep learning architectures. These advancements aim to improve prediction accuracy, particularly in under-represented regions, and enhance uncertainty quantification, leading to more robust and reliable models across diverse applications such as age estimation and cardiovascular health assessment.