ClaVis: An Interactive Visual Comparison System for Classifiers

We present a training and visualization system that enables users to visually compare many different classifiers and parameter configurations in their performance and behavior. Our approach is plugin-based and classifier-agnostic and allows users to add their own datasets and classifier implementations. It provides multiple visualizations, including a multi-variate ranking, a similarity map, a scatterplot that shows correlations between parameters and scores, as well as a training history chart. We enable analysts to interactively filter, highlight, colorize, sort, and group the displayed data. Using an iterative process, we developed our approach in cooperation with domain experts who apply machine learning for natural language processing. Our evaluation consists of two pair analytics studies and a survey with students. It demonstrates the effectiveness and usability of the implementation. Domain experts, teachers, and students showed interest in utilizing it.

Project – Article - PDF - GitHub

@inproceedings{heyen2020,
  author = {Heyen, Frank and Munz, Tanja and Neumann, Michael and Ortega, Daniel and Vu, Ngoc Thang and Weiskopf, Daniel and Sedlmair, Michael},
  title = {ClaVis: An Interactive Visual Comparison System for Classifiers},
  year = {2020},
  publisher = {Association for Computing Machinery},
  booktitle = {Proceedings of the International Conference on Advanced Visual Interfaces},
  number = {9},
  pages = {1--9},
  location = {Salerno, Italy},
  series = {AVI '20}
  doi = {10.1145/3399715.3399814},
}

Download BibTeX