CLASSEE - a Visualization Tool for Accessible Evaluation of Classification Performance
Emma Beauxis-Aussalet, J. vanDoorn, Lynda Hardman, Max Welling · Centrum Wiskunde & Informatica (CWI), the national research institute for mathematics and computer science in the Netherlands · 2016
Machine Learning techniques for automatic classification have reached a broad range of applications.But the technology transfers face issues with user trust and acceptance, as classification results inherently contain errors.Machine Learning experts rely on widely-established error measurement methods and uncertainty visualizations.However end-users are not familiar with these uncertainty visualizations, and underlying error measures.Simplified visualization designs were proposed to address this issue (Fig. 1).Machine Learning experts showed interest in using such designs to communicate with end-users.However, they wish to continue using the expert visualizations they are most familiar with.Hence we developed an interactive interface to explore classification uncertainty using visualization alternatives.We address the needs of technology providers who continuously improve classification algorithms, and communicate their performance for different application domains.We designed simple interactions for navigating through classifiers, datasets and visualizations.Our tool is developed with the D3 library, and the visualization components are delivered as open-source tools.