out-of-distribution samples

Out-of-distribution samples are data points that originate from a different distribution than the one on which the model was trained. Identifying and handling these samples is crucial for maintaining model performance in real-world applications.

6 papers