@inproceedings{johansen-socher-2017-learning,
    title = "Learning when to skim and when to read",
    author = "Johansen, Alexander  and
      Socher, Richard",
    editor = "Blunsom, Phil  and
      Bordes, Antoine  and
      Cho, Kyunghyun  and
      Cohen, Shay  and
      Dyer, Chris  and
      Grefenstette, Edward  and
      Hermann, Karl Moritz  and
      Rimell, Laura  and
      Weston, Jason  and
      Yih, Scott",
    booktitle = "Proceedings of the 2nd Workshop on Representation Learning for {NLP}",
    month = aug,
    year = "2017",
    address = "Vancouver, Canada",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/iwcs-25-ingestion/W17-2631/",
    doi = "10.18653/v1/W17-2631",
    pages = "257--264",
    abstract = "Many recent advances in deep learning for natural language processing have come at increasing computational cost, but the power of these state-of-the-art models is not needed for every example in a dataset. We demonstrate two approaches to reducing unnecessary computation in cases where a fast but weak baseline classier and a stronger, slower model are both available. Applying an AUC-based metric to the task of sentiment classification, we find significant efficiency gains with both a probability-threshold method for reducing computational cost and one that uses a secondary decision network."
}