Browsing by Author "Ngo, Quynh Quang"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Interactive Generation of 1D Embeddings from 2D Multi-dimensional Data Projections
    (The Eurographics Association, 2020) Ngo, Quynh Quang; Linsen, Lars; Krüger, Jens and Niessner, Matthias and Stückler, Jörg
    Visual analysis of multi-dimensional data is commonly supported by mapping the data to a 2D embedding. When analyzing a sequence of multi-dimensional data, e.g., in case of temporal data, the usage of 1D embeddings allows for plotting the entire sequence in a 2D layout. Despite the good performance in generating 2D embeddings, 1D embeddings often exhibit a much lower quality for pattern recognition tasks. We propose to overcome the issue by involving the user to generate 1D embeddings of multi-dimensional data in a two-step procedure: We first generate a 2D embedding and then leave the task of reducing the 2D to a 1D embedding to the user. We demonstrate that an interactive generation of 1D embeddings from 2D projected views can be performed efficiently, effectively, and targeted towards an analysis task. We compare the performance of our approach against automatically generated 1D and 2D embeddings involving a user study for our interactive approach. We test the 1D approaches when being applied to time-varying multi-dimensional data.
  • Thumbnail Image
    Item
    Visual Analytics of Simulation Ensembles for Network Dynamics
    (The Eurographics Association, 2019) Ngo, Quynh Quang; Hütt, Marc-Thorsten; Linsen, Lars; Schulz, Hans-Jörg and Teschner, Matthias and Wimmer, Michael
    A central question in the field of Network Science is to analyze the role of a given network topology on the dynamical behavior captured by time-varying simulations executed on the network. These dynamical systems are also influenced by global simulation parameters. We present a visual analytics approach that supports the investigation of the impact of the parameter settings, i.e., how parameter choices change the role of network topology on the simulations' dynamics. To answer this question, we are analyzing ensembles of simulation runs with different parameter settings executed on a given network topology. We relate the nodes' topological structures to their dynamical similarity in a 2D plot based on an interactively defined hierarchy of topological properties and a 1D embedding for the dynamical similarity. We evaluate interactively defined topological groups with respect to matching dynamical behavior, which we visually encode as graphs of the function of the considered simulation parameter. Interactive filtering and coordinated views allow for a detailed analysis of the parameter space with respect to topology-dynamics relations. Our visual analytics approach is applied to scenarios for excitable dynamics on synthetic and real brain connectome networks.
  • Thumbnail Image
    Item
    Visual Ensemble Analysis to Study the Influence of Hyper-parameters on Training Deep Neural Networks
    (The Eurographics Association, 2019) Hamid, Sagad; Derstroff, Adrian; Klemm, Sören; Ngo, Quynh Quang; Jiang, Xiaoyi; Linsen, Lars; Archambault, Daniel and Nabney, Ian and Peltonen, Jaakko
    A good deep neural network design allows for efficient training and high accuracy. The training step requires a suitable choice of several hyper-parameters. Limited knowledge exists on how the hyper-parameters impact the training process, what is the interplay of multiple hyper-parameters, and what is the interrelation of hyper-parameters and network topology. In this paper, we present a structured analysis towards these goals by investigating an ensemble of training runs.We propose a visual ensemble analysis based on hyper-parameter space visualizations, performance visualizations, and visualizing correlations of topological structures. As a proof of concept, we apply our approach to deep convolutional neural networks.