Mag. René Čutura
Full-Stack Developer & Data Visualization Specialist
Publications
ISilDR: Isometric Seriation-based Dimensionality Reduction for Visual Cluster Analysis (05/2026)
René Cutura❖ Sophie Sadler❖ Quynh Quang Ngo❖ Michaël Aupetit❖ Michael Sedlmair
Visual cluster analysis is a central task in exploring multidimensional (MD) data, typically supported by Dimensionality Reduction (DR) techniques that spatialize MD data similarities as point patterns in scatterplots. However, these layouts are often limited by unavoidable distortions—specifically false neighbors (where separated clusters appear to overlap) and missing neighbors (where true clusters are split into falsely separated groups). In this work, the authors propose Isometric Seriation-based Dimensionality Reduction (ISilDR), a family of techniques that provably generate at most missing neighbors. This contrasts with orthogonal linear projections (OLP) like Principal Component Analysis (PCA), which never produce missing neighbors but do produce false neighbors. ISilDR creates a seriation of MD data points (an ordering along a 1D projection axis) and spaces consecutive points by their MD distance. An mD ISilDR can be obtained by combining m 1D ISilDRs. The authors provide a systematic, formal analysis using epsilon-neighborhood graphs to study the theoretical and empirical characteristics of ISilDR and OLP variants. They derive rules for discovering true MD cluster patterns by interactively linking ISilDR and OLP coordinated layouts, illustrating through case studies how this combination enables more trustworthy visual cluster analysis.
Introduces ISilDR, a dimensionality reduction technique for visual cluster analysis that mathematically guarantees at most missing neighbors, preventing false neighbor distortions. Combines 1D seriation with orthogonal projections for trustworthy data exploration.
SiGrid: Gridifying Scatterplots with Sector-Based Regularization and Hagrid (11/2025)
René Cutura❖ Hennes Rave❖ Quynh Quang Ngo❖ Vladimir Molchanov❖ Lars Linsen❖ Daniel Weiskopf❖ Michael Sedlmair
Hagrid is a state-of-the-art space-filling-curve-based method for gridifying scatterplots. However, it exhibits limitations in preserving the global structures of scatterplots with areas of varying density due to the incompatibility of adapting the granularity level of the underlying space-filling curve to regions with different densities. To compensate for this shortcoming, we introduce SiGrid that combines Hagrid with the Sector-Based Regularization (SBR) technique. SiGrid applies SBR to generate a scatterplot with a more uniform and generally lower density as an intermediate step. This intermediate scatterplot can then be fed to Hagrid for improved results. We quantitatively evaluate SiGrid by comparing it to Hagrid over a set of 502 scatterplots of different sizes, ranging from 50 to 10000 points per dataset, using relevant quality metrics. While generally slower, the results demonstrate that SiGrid outperforms Hagrid regarding the quality metrics of rank-wise neighborhood preservation (trustworthiness), ordering preservation, and pairwise distance preservation (cross-correlation).
Extends Hagrid with sector-based regularization to better handle datasets with varying density, improving neighborhood preservation in large-scale, grid-based scatterplots.
Hagrid: using Hilbert and Gosper curves to gridify scatterplots (7/2022)
René Cutura❖ Cristina Morariu❖ Zhanglin Cheng❖ Yunhai Wang❖ Daniel Weiskopf❖ Michael Sedlmair
A common enhancement of scatterplots represents points as small multiples, glyphs, or thumbnail images. As this encoding often results in overlaps, a general strategy is to alter the position of the data points, for instance, to a grid-like structure. Previous approaches rely on solving expensive optimization problems or on dividing the space that alter the global structure of the scatterplot. To find a good balance between efficiency and neighborhood and layout preservation, we propose HAGRID, a technique that uses space-filling curves (SFCs) to “gridify” a scatterplot without employing expensive collision detection and handling mechanisms. Using SFCs ensures that the points are plotted close to their original position, retaining approximately the same global structure. The resulting scatterplot is mapped onto a rectangular or hexagonal grid, using Hilbert and Gosper curves. We discuss and evaluate the theoretic runtime of our approach and quantitatively compare our approach to three state-of-the-art gridifying approaches, DGRID, Small multiples with gaps SMWG, and CorrelatedMultiples CMDS, in an evaluation comprising 339 scatterplots. Here, we compute several quality measures for neighborhood preservation together with an analysis of the actual runtimes. The main results show that, compared to the best other technique, HAGRID is faster by a factor of four, while achieving similar or even better quality of the gridified layout. Due to its computational efficiency, our approach also allows novel applications of gridifying approaches in interactive settings, such as removing local overlap upon hovering over a scatterplot.
An efficient algorithm using space-filling curves to gridify scatterplots and resolve overlap without expensive collision detection, balancing speed and layout preservation.
DaRt: Generative Art using Dimensionality Reduction Algorithms (10/2021)
René Cutura❖ Katrin Angerbauer❖ Frank Heyen❖ Natalie Hube❖ Michael Sedlmair
Dimensionality Reduction (DR) is a popular technique that is often used in Machine Learning and Visualization communities to analyze high-dimensional data. The approach is empirically proven to be powerful for uncovering previously unseen structures in the data. While observing the results of the intermediate optimization steps of DR algorithms, we coincidently discovered the artistic beauty of the DR process. With enthusiasm for the beauty, we decided to look at DR from a generative art lens rather than their technical application aspects and use DR techniques to create artwork. Particularly, we use the optimization process to generate images, by drawing each intermediate step of the optimization process with some opacity over the previous intermediate result. As another alternative input, we used a neural-network model for face-landmark detection, to apply DR to portraits, while maintaining some facial properties, resulting in abstracted facial avatars. In this work, we provide such a collection of such artwork.
Uses the optimization process of dimensionality reduction as a generative art medium, producing complex visual artworks from data by overlaying intermediate algorithm steps.
Hagrid — Gridify Scatterplots with Hilbert and Gosper Curves (09/2021)
René Cutura❖ Cristina Morariu❖ Zhanglin Cheng❖ Yunhai Wang❖ Daniel Weiskopf❖ Michael Sedlmair
A common enhancement of scatterplots represents points as small multiples, glyphs, or thumbnail images. As this encoding often results in overlaps, a general strategy is to alter the position of the data points, for instance, to a grid-like structure. Previous approaches rely on solving expensive optimization problems or on dividing the space that alter the global structure of the scatterplot. To find a good balance between efficiency and neighborhood and layout preservation, we propose Hagrid, a technique that uses space-filling curves (SFCs) to “gridify” a scatterplot without employing expensive collision detection and handling mechanisms. Using SFCs ensures that the points are plotted close to their original position, retaining approximately the same global structure. The resulting scatterplot is mapped onto a rectangular or hexagonal grid, using Hilbert and Gosper curves. We discuss and evaluate the theoretic runtime of our approach and quantitatively compare our approach to three state-of-the-art gridifying approaches, DGrid, Small multiples with gaps SMWG, and CorrelatedMultiples CMDS, in an evaluation comprising 339 scatterplots. Here, we compute several quality measures for neighborhood preservation together with an analysis of the actual runtimes. The main results show that, compared to the best other technique, Hagrid is faster by a factor of four, while achieving similar or even better quality of the gridified layout. Due to its computational efficiency, our approach also allows novel applications of gridifying approaches in interactive settings, such as removing local overlap upon hovering over a scatterplot.
An efficient algorithm using space-filling curves to gridify scatterplots and resolve overlap without expensive collision detection, balancing speed and layout preservation.
funded by: FFG (ViSciPub)DruidJS — A JavaScript Library for Dimensionality Reduction (08/2020)
René Cutura❖ Christoph Kralj❖ Michael Sedlmair
Dimensionality reduction (DR) is a widely used technique for visualization. Nowadays, many of these visualizations are developed for the web, most commonly using JavaScript as the underlying programming language. So far, only few DR methods have a JavaScript implementation though, necessitating developers to write wrappers around implementations in other languages. In addition, those DR methods that exist in JavaScript libraries, such as PCA, t-SNE, and UMAP, do not offer consistent programming interfaces, hampering the quick integration of different methods. Toward a coherent and comprehensive DR programming framework, we developed an open source JavaScript library named DruidJS. Our library contains implementations of ten different DR algorithms, as well as the required linear algebra techniques, tools, and utilities.
A comprehensive open-source JavaScript library implementing over ten dimensionality reduction algorithms, enabling complex data projections to run client-side in the browser.
funded by: FFG (ViSciPub)Compadre — Comparing and Exploring High-Dimensional Data with Dimensionality Reduction Algorithms and Matrix Visualizations (02/2020)
René Cutura❖ Michael Aupetit❖ Jean-Daniel Fekete❖ Michael Sedlmair
We propose Compadre, a tool for visual analysis for comparing distances of high-dimensional (HD) data and their low-dimensional projections. At the heart is a matrix visualization to represent the discrepancy between distance matrices, linked side-by-side with 2D scatterplot projections of the data. Using different examples and datasets, we illustrate how this approach fosters (1) evaluating dimensionality reduction techniques w.r.t. how well they project the HD data, (2) comparing them to each other side-by-side, and (3) evaluate important data features through subspace comparison. We also present a case study, in which we analyze IEEE VIS authors from 1990 to 2018, and gain new insights on the relationships between coauthors, citations, and keywords. The coauthors are projected as accurately with UMAP as with t-SNE but the projections show different insights. The structure of the citation subspace is very different from the coauthor subspace. The keyword subspace is noisy yet consistent among the three IEEE VIS sub-conferences.
A visual analytics tool for comparing high-dimensional distances and low-dimensional projections using matrix visualizations, facilitating side-by-side evaluation.
funded by: FFG (ViSciPub)VisCoDeR — A tool for Visually Comparing Dimensionality Reduction Algorithms (11/2017)
Rene Cutura❖ Stefan Holzer❖ Michael Aupetit❖ Michael Sedlmair
We propose VisCoDeR, a tool that leverages comparative visualization to support learning and analyzing different dimensionality reduction (DR) methods. VisCoDeR fosters two modes. The Discover mode allows to qualitatively compare several DR results by juxtaposing and linking the resulting scatterplots. The Explore mode allows for analyzing hundreds of differently parameterized DR results in a quantitative way. We present case studies that show that our approach helps to understand similarities and differences between DR algorithms.
An interactive web tool for learning and analyzing dimensionality reduction methods through qualitative comparison of scatterplots and parameter exploration.
as co-author
WoVis: Interactive Visualization of Word Embeddings for Semantic Change in Historical and Dialectal Language Resources (05/2026)
Filip Miletić❖ Maximilian Henkel❖ René Cutura❖ Sophie Sadler❖ Quynh Quang Ngo❖ Michael Sedlmair❖ Sabine Schulte im WaldeComputational modeling of language variation and change often relies on comparisons of word embeddings induced from existing historical and dialectal language resources. However, their use in the wider linguistics research community and in application domains such as lexicography is challenged by their limited manipulability for non-technical users, which in turn exacerbates the underuse of such resources. Aiming to foster a broader uptake of embedding-based analyses, we introduce WoVis, an interactive visualization tool designed to compare word embedding models in analyses of semantic change. Our system supports simultaneous model comparisons along two dimensions (e.g., language varieties and time periods) and provides analyses at different levels of granularity: an overview of the full vocabulary across all word embedding models, distributional behavior of individual words, targeted comparisons of word pairs, and model-external lexical features such as frequency and affective norms. We illustrate the utility of our system on two languages, German and English, with analyses of word usage across language varieties as well as time: West vs. East Germany, 1950–1989; and general-domain US vs. scientific UK English, ca. 1800–2000.
Introduces WoVis, an interactive visualization tool designed to compare word embedding models along two dimensions (e.g., language varieties and time) to analyze semantic change at different levels of granularity.
funded by: DFG (TRR 161)An Image Quality Dataset with Triplet Comparisons for Multi-dimensional Scaling (06/2024)
Mohsen Jenadeleh❖ Frederik L. Dennig❖ René Cutura❖ Quynh Quang Ngo❖ Daniel A. Keim❖ Michael Sedlmair❖ Dietmar Saupe
In the early days of perceptual image quality research more than 30 years ago, the multidimensionality of distortions in perceptual space was considered important. However, research focused on scalar quality as measured by mean opinion scores. With our work, we intend to revive interest in this relevant area by presenting a first pilot dataset of annotated triplet comparisons for image quality assessment. It contains one source stimulus together with distorted versions derived from 7 distortion types at 12 levels each. Our crowdsourced and curated dataset contains roughly 50,000 responses to 7,000 triplet comparisons. We show that the multidimensional embedding of the dataset poses a challenge for many established triplet embedding algorithms. Finally, we propose a new reconstruction algorithm, dubbed logistic triplet embedding (LTE) with Tikhonov regularization. It shows promising performance. This study helps researchers to create larger datasets and better embedding techniques for multidimensional image quality.
Presents a unique dataset of 50,000 human triplet comparisons to analyze multidimensional image distortions, and proposes a new reconstruction algorithm (LTE) for human perception.
funded by: DFG (TRR 161)Predicting User Preferences of Dimensionality Reduction Embedding Quality (09/2022)
Cristina Morariu❖ Adrien Bibal❖ René Cutura❖ Benoit Frenay❖ Michael Sedlmair
A plethora of dimensionality reduction techniques have emerged over the past decades, leaving researchers and analysts with a wide variety of choices for reducing their data, all the more so given some techniques come with additional hyper-parametrization (e.g., t-SNE, UMAP, etc.). Recent studies are showing that people often use dimensionality reduction as a black-box regardless of the specific properties the method itself preserves. Hence, evaluating and comparing 2D embeddings is usually qualitatively decided, by setting embeddings side-by-side and letting human judgment decide which embedding is the best. In this work, we propose a quantitative way of evaluating embeddings, that nonetheless places human perception at the center. We run a comparative study, where we ask people to select “good” and “misleading” views between scatterplots of low-dimensional embeddings of image datasets, simulating the way people usually select embeddings. We use the study data as labels for a set of quality metrics for a supervised machine learning model whose purpose is to discover and quantify what exactly people are looking for when deciding between embeddings. With the model as a proxy for human judgments, we use it to rank embeddings on new datasets, explain why they are relevant, and quantify the degree of subjectivity when people select preferred embeddings.
A machine learning framework that predicts user preferences for 2D embeddings, using human judgments to automatically evaluate and rank projection quality.
funded by: DFG (TRR 161)Accessibility for Color Vision Deficiencies: Challenges and Findings of a Large Scale Study on Paper Figures (04/2022)
Katrin Angerbauer❖ Nils Rodrigues❖ René Cutura❖ Seyda Öney❖ Nelusa Pathmanathan❖ Cristina Morariu❖ Daniel Weiskopf❖ Michael Sedlmair
We present an exploratory study on the accessibility of images in publications when viewed with color vision deficiencies (CVDs). The study is based on 1,710 images sampled from a visualization dataset (VIS30K) over five years. We simulated four CVDs on each image. First, four researchers (one with a CVD) identified existing issues and helpful aspects in a subset of the images. Based on the resulting labels, 200 crowdworkers provided 30,000 ratings on present CVD issues in the simulated images. We analyzed this data for correlations, clusters, trends, and free text comments to gain a first overview of paper figure accessibility. Overall, about 60 % of the images were rated accessible. Furthermore, our study indicates that accessibility issues are subjective and hard to detect. On a meta-level, we reflect on our study experience to point out challenges and opportunities of large-scale accessibility studies for future research directions.
A large-scale study of over 1,700 publication figures regarding accessibility under color vision deficiencies, providing design guidelines for inclusive scientific graphics.
funded by: DFG (EXC 2075 & TRR 161)Metaphorical Visualization: Mapping Data to Familiar Concepts (04/2022)
Gleb Tkachev❖ René Cutura❖ Michael Sedlmair❖ Steffen Frey❖ Thomas Ertl
We present a new approach to visualizing data that is well-suited for personal and casual applications. The idea is to map the data to another dataset that is already familiar to the user, and then rely on their existing knowledge to illustrate relationships in the data. We construct the map by preserving pairwise distances or by maintaining relative values of specific data attributes. This metaphorical mapping is very flexible and allows us to adapt the visualization to its application and target audience. We present several examples where we map data to different domains and representations. This includes mapping data to cat images, encoding research interests with neural style transfer and representing movies as stars in the night sky. Overall, we find that although metaphors are not as accurate as the traditional techniques, they can help design engaging and personalized visualizations.
Maps data onto familiar domains (e.g., cats, stars) to increase user engagement, focusing on personalization for casual and exploratory everyday data visualization.
funded by: FFG (ViSciPub)Illegible Semantics: Exploring the Design Space of Metal Logos (09/2021)
Gerrit J. Rijken❖ René Cutura❖ Frank Heyen❖ Michael Sedlmair❖ Michael Correll❖ Jason Dykes❖ Noeska Smit
The logos of metal bands can be by turns gaudy, uncouth, or nearly illegible. Yet, these logos work: they communicate sophisticated notions of genre and emotional affect. In this paper we use the design considerations of metal logos to explore the space of “illegible semantics”: the ways that text can communicate information at the cost of readability, which is not always the most important objective. In this work, drawing on formative visualization theory, professional design expertise, and empirical assessments of a corpus ofmetal band logos, we describe a design space of metal logos and present a tool through which logo characteristics can be explored through visualization. We investigate ways in which logo designers imbue their text with meaning and consider opportunities and implications for visualization more widely.
Explores the design space of metal band logos where styling is prioritized over legibility, analyzing how text can communicate genre and emotional affect beyond readability.
funded by: DFG (TRR 161)DumbleDR: Predicting User Preferences of Dimensionality Reduction Projection Quality (05/2021)
Cristina Morariu❖ Adrien Bibal❖ René Cutura❖ Benoit Frenay❖ Michael Sedlmair
A plethora of dimensionality reduction techniques have emerged over the past decades, leaving researchers and analysts with a wide variety of choices for reducing their data, all the more so given some techniques come with additional parametrization (e.g. t-SNE, UMAP, etc.). Recent studies are showing that people often use dimensionality reduction as a black-box regardless of the specific properties the method itself preserves. Hence, evaluating and comparing 2D projections is usually qualitatively decided, by setting projections side-by-side and letting human judgment decide which projection is the best. In this work, we propose a quantitative way of evaluating projections, that nonetheless places human perception at the center. We run a comparative study, where we ask people to select 'good' and 'misleading' views between scatterplots of low-level projections of image datasets, simulating the way people usually select projections. We use the study data as labels for a set of quality metrics whose purpose is to discover and quantify what exactly people are looking for when deciding between projections. With this proxy for human judgments, we use it to rank projections on new datasets, explain why they are relevant, and quantify the degree of subjectivity in projections selected.
A machine learning framework to predict user preferences for 2D projections, automatically ranking them based on human perception.
funded by: FFGCaarvida: Visual Analytics for Test Drive Videos (10/2020)
Alexander Achberger❖ René Cutura❖ Oguzhan Türksoy❖ Michael Sedlmair
We report on an interdisciplinary visual analytics project wherein automotive engineers analyze test drive videos. These videos are annotated with navigation-specific augmented reality (AR) content, and the engineers need to identify issues and evaluate the behavior of the underlying AR navigation system. With the increasing amount of video data, traditional analysis approaches can no longer be conducted in an acceptable timeframe. To address this issue, we collaboratively developed Caarvida, a visual analytics tool that helps engineers to accomplish their tasks faster and handle an increased number of videos. Caarvida combines automatic video analysis with interactive and visual user interfaces. We conducted two case studies which show that Caarvida successfully supports domain experts and speeds up their task completion time.
A visual analytics system for automotive engineers to analyze test drive videos and AR data, speeding up troubleshooting by synchronizing video streams and sensor data.