SEQ2SEQ-VIS : A Visual Debugging Tool for Sequence-to-Sequence Models

被引:125
作者
Strobelt, Hendrik [1 ,2 ]
Gehrmann, Sebastian [3 ]
Behrisch, Michael [4 ]
Perer, Adam [1 ,2 ]
Pfister, Hanspeter [4 ]
Rush, Alexander M. [3 ]
机构
[1] IBM Res, Yorktown Hts, NY 10598 USA
[2] MIT IBM Watson Al Lab, Cambridge, MA 02142 USA
[3] Harvard NLP Grp, Cambridge, MA USA
[4] Harvard Visual Comp Grp, Cambridge, MA USA
关键词
Explainable AI; Visual Debugging; Visual Analytics; Machine Learning; Deep Learning; NLP;
D O I
10.1109/TVCG.2018.2865044
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
Neural sequence-to-sequence models have proven to be accurate and robust for many sequence prediction tasks, and have become the standard approach for automatic translation of text. The models work with a five-stage blackbox pipeline that begins with encoding a source sequence to a vector space and then decoding out to a new target sequence. This process is now standard, but like many deep learning methods remains quite difficult to understand or debug. In this work, we present a visual analysis tool that allows interaction and "what if"-style exploration of trained sequence-to-sequence models through each stage of the translation process. The aim is to identify which patterns have been learned. to detect model errors, and to probe the model with counterfactual scenario. We demonstrate the utility of our tool through several real-world sequence-to-sequence use cases on large-scale models.
引用
收藏
页码:353 / 363
页数:11
相关论文
共 55 条
  • [11] [Anonymous], J OPEN RES SOFTWARE
  • [12] [Anonymous], 2016, Abstractive text summarization using sequence-tosequence rnns and beyond
  • [13] [Anonymous], 2016, P C ASS MACH TRANSL
  • [14] [Anonymous], ARXIV E PRINTS
  • [15] [Anonymous], DISTILL
  • [16] [Anonymous], 2015, COMPUTER SCI
  • [17] [Anonymous], 2016, Understanding neural networks through representation erasure
  • [18] [Anonymous], GUARDIAN
  • [19] [Anonymous], 2017, ARXIV171010777
  • [20] [Anonymous], 2015, ARXIV150900685, DOI DOI 10.18653/V1/D15-1044