You can't handle the (dirty) truth: Data-centric insights improve pseudo-labeling
This addresses the overlooked issue of labeled data quality in semi-supervised learning for practitioners dealing with noisy real-world data, representing an incremental improvement.
The paper tackles the problem that pseudo-labeling methods assume labeled data is perfect, which is often violated in practice with issues like mislabeling. They introduce DIPS, a data characterization and selection framework that improves various pseudo-labeling methods across real-world datasets, enhancing data efficiency and reducing performance gaps between methods.
Pseudo-labeling is a popular semi-supervised learning technique to leverage unlabeled data when labeled samples are scarce. The generation and selection of pseudo-labels heavily rely on labeled data. Existing approaches implicitly assume that the labeled data is gold standard and 'perfect'. However, this can be violated in reality with issues such as mislabeling or ambiguity. We address this overlooked aspect and show the importance of investigating labeled data quality to improve any pseudo-labeling method. Specifically, we introduce a novel data characterization and selection framework called DIPS to extend pseudo-labeling. We select useful labeled and pseudo-labeled samples via analysis of learning dynamics. We demonstrate the applicability and impact of DIPS for various pseudo-labeling methods across an extensive range of real-world tabular and image datasets. Additionally, DIPS improves data efficiency and reduces the performance distinctions between different pseudo-labelers. Overall, we highlight the significant benefits of a data-centric rethinking of pseudo-labeling in real-world settings.