Browsing by Author "Kerren, Andreas"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    SimBaTex: Similarity-based Text Exploration
    (The Eurographics Association, 2021) Witschard, Daniel; Jusufi, Ilir; Kerren, Andreas; Byška, Jan and Jänicke, Stefan and Schmidt, Johanna
    Natural language processing in combination with visualization can provide efficient ways to discover latent patterns of similarity which can be useful for exploring large sets of text documents. In this poster abstract, we describe the ongoing work on a visual analytics application, called SimBaTex, which is based on embedding technology, dynamic specification of similarity criteria, and a novel approach for similarity-based clustering. The goal of SimBaTex is to provide search-and-explore functionality to enable the user to identify items of interest in a large set of text documents by interactive assessment of both high-level similarity patterns and pairwise similarity of chosen texts.
  • Thumbnail Image
    Item
    Text Visualization Revisited: The State of the Field in 2019
    (The Eurographics Association, 2019) Kucher, Kostiantyn; Kerren, Andreas; Madeiras Pereira, João and Raidou, Renata Georgia
    Text and document data visualization is an important research field within information visualization and visual analytics with multiple application domains including digital humanities and social media, for instance. During the past five years, we have been collecting text visualization techniques described in peer-reviewed literature, categorizing them according to a detailed categorization schema, and providing the resulting manually curated collection in an online survey browser. In this poster paper, we present the updated results of analyses of this data set as of spring 2019. Compared to the recent surveys and meta-analyses that mainly focus on particular aspects and problems related to text visualization, our results provide an overview of the current state of the text visualization field and the respective research community in general.
  • Thumbnail Image
    Item
    VisEvol: Visual Analytics to Support Hyperparameter Search through Evolutionary Optimization
    (The Eurographics Association and John Wiley & Sons Ltd., 2021) Chatzimparmpas, Angelos; Martins, Rafael M.; Kucher, Kostiantyn; Kerren, Andreas; Borgo, Rita and Marai, G. Elisabeta and Landesberger, Tatiana von
    During the training phase of machine learning (ML) models, it is usually necessary to configure several hyperparameters. This process is computationally intensive and requires an extensive search to infer the best hyperparameter set for the given problem. The challenge is exacerbated by the fact that most ML models are complex internally, and training involves trial-and-error processes that could remarkably affect the predictive result. Moreover, each hyperparameter of an ML algorithm is potentially intertwined with the others, and changing it might result in unforeseeable impacts on the remaining hyperparameters. Evolutionary optimization is a promising method to try and address those issues. According to this method, performant models are stored, while the remainder are improved through crossover and mutation processes inspired by genetic algorithms. We present VisEvol, a visual analytics tool that supports interactive exploration of hyperparameters and intervention in this evolutionary procedure. In summary, our proposed tool helps the user to generate new models through evolution and eventually explore powerful hyperparameter combinations in diverse regions of the extensive hyperparameter space. The outcome is a voting ensemble (with equal rights) that boosts the final predictive performance. The utility and applicability of VisEvol are demonstrated with two use cases and interviews with ML experts who evaluated the effectiveness of the tool.