Transfer learning is beneficial by allowing the expressive features of models pretrained on large-scale datasets to be finetuned for the target task of smaller, more domain-specific datasets. However, there is a concern that these pretrained models may come with their own biases which would propagate into the finetuned model. In this work, we investigate bias when conceptualized as both spurious correlations between the target task and a sensitive attribute as well as underrepresentation of a particular group in the dataset. Under both notions of bias, we find that (1) models finetuned on top of pretrained models can indeed inherit their biases, but (2) this bias can be corrected for through relatively minor interventions to the finetuning dataset, and often with a negligible impact to performance. Our findings imply that careful curation of the finetuning dataset is important for reducing biases on a downstream task, and doing so can even compensate for bias in the pretrained model.
翻译:转移学习的好处是,允许在大规模数据集方面预先培训的模型的清晰特征被微调,以适应更小、更特定领域数据集的目标任务,从而有利于转移学习。然而,人们担心这些预先培训的模型可能会产生自己的偏差,而这种偏差会传播到微调模型中。在这项工作中,当我们把偏差概念概念化为目标任务与敏感属性之间的虚假关联以及特定群体在数据集中的代表性不足时,我们调查偏差。 在两种偏差概念下,我们发现(1) 在经过培训的模型上方微调的模型确实可以继承其偏差,但(2) 通过微调数据集的相对较小的干预,这种偏差是可以纠正的,而且往往对业绩产生微不足道的影响。 我们的调查结果表明,仔细调整数据集对于减少下游任务的偏差很重要,这样做甚至可以弥补事先培训模型中的偏差。</s>