A Data-Based Perspective on Transfer Learning
This work addresses transfer learning brittleness for practitioners by enabling better data curation, though it is incremental in refining existing methods.
The paper tackles the problem of source dataset composition in transfer learning, showing that removing detrimental datapoints identified by their framework improves performance on various target tasks, with concrete gains demonstrated from ImageNet.
It is commonly believed that in transfer learning including more pre-training data translates into better performance. However, recent evidence suggests that removing data from the source dataset can actually help too. In this work, we take a closer look at the role of the source dataset's composition in transfer learning and present a framework for probing its impact on downstream performance. Our framework gives rise to new capabilities such as pinpointing transfer learning brittleness as well as detecting pathologies such as data-leakage and the presence of misleading examples in the source dataset. In particular, we demonstrate that removing detrimental datapoints identified by our framework improves transfer learning performance from ImageNet on a variety of target tasks. Code is available at https://github.com/MadryLab/data-transfer