Assessing the Impact of Sequence Length Learning on Classification Tasks for Transformer Encoder Models
This addresses a data bias issue in critical applications like medicine and insurance, but it is incremental as it focuses on exposing and mitigating an existing problem.
The paper tackles the sequence length learning problem in Transformer encoder models for classification, where models exploit length distributions as predictive features instead of textual information, and presents approaches to minimize its impacts, though no concrete numbers are provided.
Classification algorithms using Transformer architectures can be affected by the sequence length learning problem whenever observations from different classes have a different length distribution. This problem causes models to use sequence length as a predictive feature instead of relying on important textual information. Although most public datasets are not affected by this problem, privately owned corpora for fields such as medicine and insurance may carry this data bias. The exploitation of this sequence length feature poses challenges throughout the value chain as these machine learning models can be used in critical applications. In this paper, we empirically expose this problem and present approaches to minimize its impacts.