Learning With Noisy Labels

Learning with noisy labels refers to the task where, in the training data, some labels are maliciously altered, causing errors in labels that were originally from a clean distribution. The goal of this task is to design and develop algorithms that can effectively identify and correct these erroneous labels under such suboptimal data conditions, thereby enhancing the robustness and generalization ability of the model. Learning with noisy labels not only has significant application value in computer vision but can also be widely applied to other machine learning tasks, improving the adaptability and reliability of models in real-world scenarios.