• Publikationen
Show publication details

Bernard, Jürgen; Hutter, Marco; Reinemuth, Heiko; Pfeifer, Hendrik; Bors, Christian; Kohlhammer, Jörn

Visual-Interactive Preprocessing of Multivariate Time Series Data


Computer Graphics Forum

Eurographics Conference on Visualization (EuroVis) <21, 2019, Porto, Portugal>

Pre-processing is a prerequisite to conduct effective and efficient downstream data analysis. Pre-processing pipelines often require multiple routines to address data quality challenges and to bring the data into a usable form. For both the construction and the refinement of pre-processing pipelines, human-in-the-loop approaches are highly beneficial. This particularly applies to multivariate time series, a complex data type with multiple values developing over time. Due to the high specificity of this domain, it has not been subject to in-depth research in visual analytics. We present a visual-interactive approach for preprocessing multivariate time series data with the following aspects. Our approach supports analysts to carry out six core analysis tasks related to pre-processing of multivariate time series. To support these tasks, we identify requirements to baseline toolkits that may help practitioners in their choice. We characterize the space of visualization designs for uncertainty-aware pre-processing and justify our decisions. Two usage scenarios demonstrate applicability of our approach, design choices, and uncertainty visualizations for the six analysis tasks. This work is one step towards strengthening the visual analytics support for data pre-processing in general and for uncertainty-aware pre-processing of multivariate time series in particular.

Show publication details

Bernard, Jürgen; Hutter, Marco; Zeppelzauer, Matthias; Fellner, Dieter W.; Sedlmair, Michael

Comparing Visual-Interactive Labeling with Active Learning: An Experimental Study


IEEE Transactions on Visualization and Computer Graphics

Labeling data instances is an important task in machine learning and visual analytics. Both fields provide a broad set of labeling strategies, whereby machine learning (and in particular active learning) follows a rather model-centered approach and visual analytics employs rather user-centered approaches (visual-interactive labeling). Both approaches have individual strengths and weaknesses. In this work, we conduct an experiment with three parts to assess and compare the performance of these different labeling strategies. In our study, we (1) identify different visual labeling strategies for user-centered labeling, (2) investigate strengths and weaknesses of labeling strategies for different labeling tasks and task complexities, and (3) shed light on the effect of using different visual encodings to guide the visual-interactive labeling process. We further compare labeling of single versus multiple instances at a time, and quantify the impact on efficiency. We systematically compare the performance of visual interactive labeling with that of active learning. Our main findings are that visual-interactive labeling can outperform active learning, given the condition that dimension reduction separates well the class distributions. Moreover, using dimension reduction in combination with additional visual encodings that expose the internal state of the learning model turns out to improve the performance of visual-interactive labeling.

Show publication details

Steiger, Martin; Bernard, Jürgen; Thum, Simon; Mittelstädt, Sebastian; Hutter, Marco; Keim, Daniel A.; Kohlhammer, Jörn

Explorative Analysis of 2D Color Maps


WSCG 2015. Full Papers Proceedings

International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision (WSCG) <23, 2015, Plzen, Czech Republic>

Color is one of the most important visual variables in information visualization. In many cases, two-dimensional information can be color-coded based on a 2D color map. A variety of color maps as well as a number of quality criteria for the use of color have been presented. The choice of the best color map depends on the analytical task users intend to perform and the design space in choosing an appropriate 2D color map is large. In this paper, we present the ColorMap-Explorer, a visual-interactive system that helps users in selecting the most appropriate 2D color map for their particular use case. ColorMap-Explorer also provides a library of many color map implementations that have been proposed in the scientific literature. To analyze their usefulness for different tasks, ColorMap-Explorer provides use case scenarios to allow users to obtain qualitative feedback. In addition, quantitative metrics are provided on a global (i.e. per color map) and local (i.e. per point) scale. ColorMap-Explorer enables users to explore the strengths and weaknesses of existing as well as user-provided color maps to find the best fit for their task. Any color map can be exported to be reused in other visualization tools. The code is published as open source software, so that the visualization community can use both the color map library and the ColorMap-Explorer tool. This also allows users to contribute new implementations.

Show publication details

Twellmeyer, James; Hutter, Marco; Behrisch, Michael; Kohlhammer, Jörn; Schreck, Tobias

The Visual Exploration of Aggregate Similarity for Multi-dimensional Clustering


IVAPP 2015. Proceedings

International Conference on Information Visualization Theory and Applications (IVAPP) <6, 2015, Berlin, Germany>

We present a visualisation prototype for the support of a novel approach to clustering called TRIAGE. TRIAGE uses aggregation functions which are more adaptable and flexible than the weighted mean for similarity modelling. While TRIAGE has proven itself in practice, the use of complex similarity models makes the interpretation of TRIAGE clusterings challenging. We address this challenge by providing analysts with a linked, matrix-based visualisation of all relevant data attributes. We employ data sampling and matrix seriation to support both effective overviews and fluid, interactive exploration using the same visual metaphor for heterogeneous attributes. The usability of our prototype is demonstrated and assessed with the help of real-world usage scenarios from the cyber-security domain.

Show publication details

Steiger, Martin; Hutter, Marco; Schader, Philipp; Kohlhammer, Jörn; Kuijper, Arjan

Exploring Simulation in Sensor Network Models


VMV 2014

Workshop on Vision, Modeling, and Visualization (VMV) <19, 2014, Darmstadt, Germany>

Simulation is an important measure to estimate different properties of a planned network such as throughput and cost. However, many parameters need to be adjusted to approximate real-world conditions properly. In this paper we present a visualization system that visually supports and guides the analysis of (physical) network simulation problems. Automatic optimizers run as a black box giving an (locally) optimal result in terms of the underlying simulation model and parameter configuration. This is often not ideal for practical usage. Our system assists the user in the process of comparing different simulations to quickly achieve the optimal configuration in terms of user preference. It highlights differences between simulation runs and indicates which parameter modification leads to the best improvement. We expect that this results in large time savings for the domain expert while configuring the simulation system.

Show publication details

Nazemi, Kawa; Kuijper, Arjan; Hutter, Marco; Kohlhammer, Jörn; Fellner, Dieter W.

Measuring Context Relevance for Adaptive Semantics Visualizations


i-KNOW 2014

International Conference on Knowledge Technologies and Data-driven Business (I-KNOW) <14, 2014, Graz, Austria>

Semantics visualizations enable the acquisition of information to amplify the acquisition of knowledge. The dramatic increase of semantics in form of Linked Data and Linked-Open Data yield search databases that allow to visualize the entire context of search results. The visualization of this semantic context enables one to gather more information at once, but the complex structures may as well confuse and frustrate users. To overcome the problems, adaptive visualizations already provide some useful methods to adapt the visualization on users' demands and skills. Although these methods are very promising, these systems do not investigate the relevance of semantic neighboring entities that commonly build most information value. We introduce two new measurements for the relevance of neighboring entities: The Inverse Instance Frequency allows weighting the relevance of semantic concepts based on the number of their instances. The Direct Relation Frequency inverse Relations Frequency measures the relevance of neighboring instances by the type of semantic relations. Both measurements provide a weighting of neighboring entities of a selected semantic instance, and enable an adaptation of retinal variables for the visualized graph. The algorithms can easily be integrated into adaptive visualizations and enhance them with the relevance measurement of neighboring semantic entities. We give a detailed description of the algorithms to enable a replication for the adaptive and semantics visualization community. With our method, one can now easily derive the relevance of neighboring semantic entities of selected instances, and thus gain more information at once, without confusing and frustrating users.

Show publication details

Hutter, Marco; Knuth, Martin; Kuijper, Arjan

Mesh Partitioning for Parallel Garment Simulation


WSCG 2014. Communication Papers Proceedings

International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision (WSCG) <22, 2014, Plzen, Czech Republic>

We present a method for partitioning meshes that allows a simple and efficient parallel implementation of different simulation methods. It is based on a generalization of the concept of independent sets from graph theory to sets of simulation elements. The general description makes it versatile and flexibly applicable in existing simulation systems. Every simulation method that formerly worked by sequentially processing a set of simulation elements can now be parallelized by partitioning the underlying set, without affecting the behavior of the simulated model.

Show publication details

Bernard, Jürgen; Hutter, Marco; Sessler, David; Schreck, Tobias; Behrisch, Michael; Kohlhammer, Jörn

Towards a User-Defined Visual-Interactive Definition of Similarity Functions for Mixed Data


IEEE Conference on Visual Analytics Science and Technology. Proceedings

IEEE Symposium on Visual Analytics Science and Technology (VAST) <9, 2014, Paris, France>

The creation of similarity functions based on visual-interactive user feedback is a promising means to capture the mental similarity notion in the heads of domain experts. In particular, concepts exist where users arrange multivariate data objects on a 2D data landscape in order to learn new similarity functions. While systems that incorporate numerical data attributes have been presented in the past, the remaining overall goal may be to develop systems also for mixed data sets. In this work, we present a feedback model for categorical data which can be used alongside of numerical feedback models in future.

Show publication details

Davey, James; Hutter, Marco; May, Thorsten; Kohlhammer, Jörn

Combining Linked Matrix Visualizations with Multi-Dimensional Clustering to Detect Cybercrime


International Workshop on Visual Analytics (EuroVA) <3, 2012, Vienna, Austria>

Our goal is to find a useful multi-dimensional clustering (MDC) of a data set from the network security domain. Algorithms from the Multi-Criteria Decision Analysis (MCDA) domain are used for the calculation of fused similarity matrices, in which the information from several dimensions are combined. However, the results of MCDA algorithms are often difficult to interpret. We present a linked, interactive, matrix-based visualization which simplifies the comparison of clusters and increases the understanding of MCDA results.

Show publication details

Lorek, Jens; Hutter, Marco [Betreuer]

Entwicklung intuitiver Navigationsmöglichkeiten für eine Virtual Prototyping Umgebung


Darmstadt, TU, Bachelor Thesis, 2008

Im Rahmen dieser Bachelorarbeit wurde das bestehende Interaktionsmodell einer Simulationssoftware an die Bedürfnisse des Virtual Prototyping angepasst und um neue Funktionen erweitert. Dabei wurde ein Kameramodell entwickelt, das bei allen Zoomstufen frei und intuitiv mit der Maus steuerbar ist. Zusätzlich wurde Möglichkeit geboten, interaktiv ausgewählte Bereiche mit Hilfe von Kamerafahrten ins Blickfeld zu rücken.

Show publication details

Hutter, Marco; Fuhrmann, Arnulph

Optimized Continuous Collision Detection for Deformable Triangle Meshes


Journal of WSCG Vol. 15 No. 1-3, 2007. Proceedings

International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision (WSCG) <15, 2007, Plzen, Czech Republic>

We present different approaches for accelerating the process of continuous collision detection for deformable triangle meshes. The main focus is upon the collision detection for simulated virtual clothing, especially for situations involving a high number of contact points between the triangle meshes, such as multi-layered garments. We show how the culling efficiency of bounding volume hierarchies may be increased by introducing additional bounding volumes for edges and vertices of the triangle mesh. We present optimized formulas for computing the time of collision for these primitives analytically, and describe an efficient iterative scheme that ensures that all collisions are treated in the correct chronological order.

Show publication details

Hutter, Marco; Fuhrmann, Arnulph [Betreuer]

Kollisionserkennung für mehrlagige Bekleidung


Darmstadt, TU, Diplomarbeit, 2005

In dieser Diplomarbeit werden verschiedene Verfahren zur Kollisionserkennung für mehrlagige Bekleidung vorgestellt. Die Kleidungsstücke sind als ein Dreiecksnetz gegeben, und werden mit Hilfe eines Masse-Feder-Modells simuliert. Bekannte Methoden zum Beschleunigen der Kollisionserkennung werden verglichen, insbesondere verschiedene Formen von Bounding Volume Hierarchies (BVH). Es wird gezeigt, wie die Effizienz von BVH zum Ausschließen nicht kollidierender Primitivpaare erhöht werden kann, indem man zusätzliche Bounding Volumes für die Knoten und Kanten des Dreiecksgitters einführt. Das Hauptaugenmerk dieser Arbeit ist die robuste kontinuierliche Kollisionserkennung, mit deren Hilfe Überschneidungen zwischen Dreiecksnetzen verhindert werden, unabhängig von der Geschwindigkeit ihrer Bewegung und der Größe der Zeitschritte, die für die numerische Integration verwendet wird. Es wird ein iteratives Verfahren vorgestellt, mit dem gewährleistet ist, dass alle Kollisionen in der richtigen zeitlichen Reihenfolge behandelt werden. Zusätzlich wird eine Kollisionsantwort für beliebige, deformierbare Dreiecksnetze beschrieben. Es werden verschiedene Methoden verglichen, mit deren Hilfe Überschneidungen zwischen Dreiecksnetzen behandelt werden können, die durch externe Constraints erzwungen werden, oder aufgrund von unzulässigen Anfangszuständen auftreten. Für mehrlagige Bekleidung können zusätzlich semantische Informationen über die Bekleidungsreihenfolge und damit über die Lage der Kleidungsstücke in Bezug auf den Avatar vorliegen. Es wird gezeigt, wie diese Informationen verwendet werden können, um unzulässige Zustände zu korrigieren.