Search

Embeddings are powerful tools for transforming complex and unstructured data into numeric formats suitable for computational analysis tasks. In this paper, we extend our previous work on using multiple embeddings for text similarity calculations to the field of networks. The embedding ensemble approach improves network reconstruction performance compared to single-embedding strategies. Our visual analytics methodology is successful in handling both text and network data, which demonstrates its generalizability beyond its originally presented scope.

Stacking (or stacked generalization) is an ensemble learning method with one main distinctiveness from the rest: even though several base models are trained on the original data set, their predictions are further used as input data for one or more metamodels arranged in at least one extra layer. Composing a stack of models can produce high-performance outcomes, but it usually involves a trial-and-error process. Therefore, our previously developed visual analytics system, StackGenVis, was mainly designed to assist users in choosing a set of top-performing and diverse models by measuring their predictive performance. However, it only employs a single logistic regression metamodel. In this paper, we investigate the impact of alternative metamodels on the performance of stacking ensembles using a novel visualization tool, called MetaStackVis. Our interactive tool helps users to visually explore different singular and pairs of metamodels according to their predictive probabilities and multiple validation metrics, as well as their ability to predict specific prob- lematic data instances. MetaStackVis was evaluated with a usage scenario based on a medical data set and via expert interviews.

Identifying sociolinguistic attributes of inter-community interactions is essential for understanding the polarization of social network communities. A wide range of computational text and network analysis methods may be applicable for this task, however, interpretation of the respective results and investigation of particularly interesting cases and subnetworks are difficult due to the scale and complexity of the data, e.g., for the Reddit platform. In this poster paper, we present an interactive visual analysis interface that facilitates network exploration and comparison at different topological and multivariate attribute scales. Users are able to investigate text- and network-based properties of social network community interactions, identify anomalies of conflict starters, or gain insight into multivariate anomalies behind groups of negative social media posts.

We describe a benchmark data generator for tracking methods for two- and three-dimensional time-dependent scalar fields. More and more topology-based tracking methods are presented in the visualization community, but the validation and evaluation of the tracking results are currently limited to qualitative visual approaches. We present a pipeline for creating different ground truth features that support evaluating tracking methods based on quantitative measures. In short, our approach randomly simulates a temporal point cloud with birth, death, split, merge, and continuation events, where the points are then used to derive a scalar field whose topological features correspond to the points. These scalar fields can be used as the input for different tracking methods, where the computed tracks can be compared against the ground truth feature evolution. This approach facilitates directly comparing the results of different tracking methods, independent of the initial feature characterization., Funding Agencies|SeRC (Swedish e-Science Research Center); ELLIIT environment for strategic research in Sweden; Swedish Research Council (VR) [2019-05487]

The problem of tracking and visualizing cyclones is still an active area of climate research, since the nature of cyclones varies depending on geospatial location and temporal season, resulting in no clear mathematical definition. Thus, many cyclone tracking methods are tailored to specific datasets and therefore do not support general cyclone extraction across the globe. To address this challenge, we present a conceptual application for exploring cyclone evolution by organizing the extracted cyclone tracks into hierarchical groups. Our approach is based on extrema tracking, and the resulting tracks can be defined in a multi-scale structure by grouping the points based on a novel feature descriptor defined on the merge tree, so-called crown features. Consequently, multiple parameter settings can be visualized and explored in a level-of-detail approach, supporting experts to quickly gain insights on cyclonic formation and evolution. We describe a general cyclone exploration pipeline that consists of four modular building blocks: (1) an extrema tracking method, (2) multiple definitions of cyclones as groups of extrema, including crown features, (3) the correlation of cyclones based on the underlying tracking information, and (4) a hierarchical visualization of the resulting feature tracks and their spatial embedding, allowing exploration on a global and local scale. In order to be as flexible as possible, our pipeline allows for exchanging every module with different techniques, such as other tracking methods and cyclone definitions., Funding Agencies|SeRC (Swedish e-Science Research Center); ELLIIT environment for strategic research in Sweden; Swedish Research Council (VR) [2019-05487]

We present a method for detecting patterns in time-varying functional magnetic resonance imaging (fMRI) data based on topological analysis. The oxygenated blood flow measured by fMRI is widely used as an indicator of brain activity. The signal is, however, prone to noise from various sources. Random brain activity, physiological noise, and noise from the scanner can reach a strength comparable to the signal itself. Thus, extracting the underlying signal is a challenging process typically approached by applying statistical methods. The goal of this work is to investigate the possibilities of recovering information from the signal using topological feature vectors directly based on the raw signal without medical domain priors. We utilize merge trees to define a robust feature vector capturing key features within a time step of fMRI data. We demonstrate how such a concise feature vector representation can be utilized for exploring the temporal development of brain activations, connectivity between these activations, and their relation to cognitive tasks., Funding Agencies|Wallenberg AI, Autonomous Systems and Software Program (WASP) - Knut and Alice Wallenberg Foundation; SeRC (Swedish e-Science Research Center); ELLIIT environment for strategic research in Sweden; Swedish Research Council (VR) [2019-05487]

We present a new topological connection method for the local bi-linear computation of Jacobi sets that improves the visual representation while preserving the topological structure and geometric configuration. To this end, the topological structure of the local bilinear method is utilized, which is given by the nerve complex of the traditional piecewise linear method. Since the nerve complex consists of higher-dimensional simplices, the local bilinear method (visually represented by the 1-skeleton of the nerve complex) leads to clutter via crossings of line segments. Therefore, we propose a homotopy-equivalent representation that uses different collapses and edge contractions to remove such artifacts. Our new connectivity method is easy to implement, comes with only little overhead, and results in a less cluttered representation., Funding Agencies|Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) [DFG 270852890-GRK 2160/2, DFG 251654672-TRR 161]; Swedish Research Council (VR) [2019-05487]; U.S. Department of Energy (DOE) [DOE DE-SC0021015]; National Science Foundation (NSF) [NSF IIS-1910733]

Stacked generalization (also called stacking) is an ensemble method in machine learning that deploys a metamodel to summarize the predictive results of heterogeneous base models organized into one or more layers. Despite being capable of producing high-performance results, building a stack of models can be a trial-and-error procedure. Thus, our previously developed visual analytics system, entitled StackGenVis, was designed to monitor and control the entire stacking process visually. In this work, we present the results of a comparative user study we performed for evaluating the StackGenVis system. We divided the study participants into two groups to test the usability and effectiveness of StackGenVis compared to Orange Visual Stacking (OVS) in an exploratory usage scenario using healthcare data. The results indicate that StackGenVis is significantly more powerful than OVS based on the qualitative feedback provided by the participants. However, the average completion time for all tasks was comparable between both tools., Funding: ELLIIT environment for strategic research in Sweden