Class Probability

Class probability estimation focuses on accurately determining the likelihood of a data point belonging to each class within a classification problem. Current research emphasizes improving the reliability and robustness of these probability estimates, particularly addressing challenges posed by imbalanced datasets, adversarial attacks, and the need for uncertainty quantification. This involves developing novel algorithms and adapting existing models like Support Vector Machines and deep neural networks, often incorporating techniques like quantile-based approaches and incorporating both data features and class probabilities for improved performance. Accurate class probabilities are crucial for various applications, including decision-making under uncertainty, model evaluation, and improving the fairness and trustworthiness of machine learning systems.