@inproceedings{zhou-srikumar-2023-metaprobe,
    title = "{METAPROBE}: A Representation- and Task-Agnostic Probe",
    author = "Zhou, Yichu  and
      Srikumar, Vivek",
    editor = "Belinkov, Yonatan  and
      Hao, Sophie  and
      Jumelet, Jaap  and
      Kim, Najoung  and
      McCarthy, Arya  and
      Mohebbi, Hosein",
    booktitle = "Proceedings of the 6th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP",
    month = dec,
    year = "2023",
    address = "Singapore",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/fix-sig-urls/2023.blackboxnlp-1.18/",
    doi = "10.18653/v1/2023.blackboxnlp-1.18",
    pages = "233--249",
    abstract = "Probing contextualized representations typically involves comparing task-specific model predictions against ground truth linguistic labels. Although this methodology shows \textit{what} information can be recovered by a classifier, it does not reveal \textit{how} a classifier uses the representation to make its decision. To address the latter problem, we ask: Do task-classifiers rely on representation- and task-independent geometric patterns in the embedding space? We explore this question by developing MetaProbe, an approach that uses geometric properties of representations to predict the behavior of task-specific classifiers (i.e., their predictions as opposed to the ground truth). Our experiments reveal the existence of universal geometric patterns across representations that can predict classifier predictions. Consequently, this allows us to posit a geometric explanation for the impressive performance of contextualized representations."
}