Class Outlier

Class outliers, data points significantly different from the majority within a specific class, pose a significant challenge for machine learning models, impacting both accuracy and explainability. Current research focuses on developing robust outlier detection methods, often employing contrastive learning, ensemble techniques combining uncertainty measures with outlier class predictions, and diverse sampling strategies to improve model generalization and reduce overconfidence in out-of-distribution predictions. These advancements are crucial for improving the reliability and safety of machine learning systems in various applications, particularly in safety-critical domains like visual recognition and open-set semi-supervised learning.