EuroVA17

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/2631436

Browse

Browsing EuroVA17 by Subject "Active learning"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Combining Cluster and Outlier Analysis with Visual Analytics
    (The Eurographics Association, 2017) Bernard, Jürgen; Dobermann, Eduard; Sedlmair, Michael; Fellner, Dieter W.; Michael Sedlmair and Christian Tominski
    Cluster and outlier analysis are two important tasks. Due to their nature these tasks seem to be opposed to each other, i.e., data objects either belong to a cluster structure or a sparsely populated outlier region. In this work, we present a visual analytics tool that allows the combined analysis of clusters and outliers. Users can add multiple clustering and outlier analysis algorithms, compare results visually, and combine the algorithms' results. The usefulness of the combined analysis is demonstrated using the example of labeling unknown data sets. The usage scenario also shows that identified clusters and outliers can share joint areas of the data space.
  • Thumbnail Image
    Item
    A Unified Process for Visual-Interactive Labeling
    (The Eurographics Association, 2017) Bernard, Jürgen; Zeppelzauer, Matthias; Sedlmair, Michael; Aigner, Wolfgang; Michael Sedlmair and Christian Tominski
    Assigning labels to data instances is a prerequisite for many machine learning tasks. Similarly, labeling is applied in visualinteractive analysis approaches. However, the strategies for creating labels often differ in the two fields. In this paper, we study the process of labeling data instances with the user in the loop, from both the machine learning and visual-interactive perspective. Based on a review of differences and commonalities, we propose the 'Visual-Interactive Labeling' (VIAL) process, conflating the strengths of both. We describe the six major steps of the process and highlight their related challenges.