Andreas Hinterreiter

Andreas Hinterreiter, Christina Humer, Bernhard Kainz et al.

ParaDime is a framework for parametric dimensionality reduction (DR). In parametric DR, neural networks are trained to embed high-dimensional data items in a low-dimensional space while minimizing an objective function. ParaDime builds on the idea that the objective functions of several modern DR techniques result from transformed inter-item relationships. It provides a common interface for specifying these relations and transformations and for defining how they are used within the losses that govern the training process. Through this interface, ParaDime unifies parametric versions of DR techniques such as metric MDS, t-SNE, and UMAP. It allows users to fully customize all aspects of the DR process. We show how this ease of customization makes ParaDime suitable for experimenting with interesting techniques such as hybrid classification/embedding models and supervised DR. This way, ParaDime opens up new possibilities for visualizing high-dimensional data.

Amal Alnouri, Andreas Hinterreiter, Christina Humer et al.

Large language model (LLM) outputs arise from complex interactions among prompts, system instructions, model parameters, and architecture. We refer to specific configurations of these factors as generation conditions, each of which can bias outputs in various ways. Understanding how different generation conditions shape model behaviors is essential for tasks such as prompt design and model evaluation, yet it remains challenging due to the stochastic and open-ended nature of text generation. We present an approach to visually compare LLM outputs across generation conditions by modeling responses as collections of linguistic choices, including content, expression, and structure. We extract these choices using natural language processing pipelines and represent their distributions across repeated samples. We then visualize these distributions as visual fingerprints, enabling direct, distribution-level comparison of condition-specific tendencies. Through four usage scenarios, we demonstrate how visual fingerprints reveal consistent patterns in LLM behavior that are difficult to observe through individual responses or aggregate metrics.

Jakob Zethofer, Andreas Hinterreiter, Lukas Schiefermüller et al.

We introduce EventColumn, a new column type that integrates event-sequence data with heterogeneous tabular attributes into a single unified table. EventColumn lets analysts compare event sequences alongside numerical, categorical, and temporal attributes at both instance and group levels, offering a compressed overview, heatmap group summaries, alignment by event types, and boxplots of similar historical items. We developed EventColumn together with collaborators from the steel industry to facilitate the analysis of production events and warehouse logistics, but the solution generalizes to a wide range of event sequence datasets with additional tabular attributes. Unlike most existing approaches that compare either event sequences or tables, EventColumn supports simultaneous comparison of both. We demonstrate its integration with Taggle and Microsoft Power BI on data from steel production logistics and on a public e-commerce dataset.

Florian Holeczek, Andreas Hinterreiter, Alex Hernandez-Garcia et al.

We present GFlowState, a visual analytics system designed to illuminate the training process of Generative Flow Networks (GFlowNets or GFNs). GFlowNets are a probabilistic framework for generating samples proportionally to a reward function. While GFlowNets have proved to be powerful tools in applications such as molecule and material discovery, their training dynamics remain difficult to interpret. Standard machine learning tools allow metric tracking but do not reveal how models explore the sample space, construct sample trajectories, or shift sampling probabilities during training. Our solution, GFlowState, allows users to analyze sampling trajectories, compare the sample space relative to reference datasets, and analyze the training dynamics. To this end, we introduce multiple views, including a chart of candidate rankings, a state projection, a node-link diagram of the trajectory network, and a transition heatmap. These visualizations enable GFlowNet developers and users to investigate sampling behavior and policy evolution, and to identify underexplored regions and sources of training failure. Case studies demonstrate how the system supports debugging and assessing the quality of GFlowNets across application domains. By making the structural dynamics of GFlowNets observable, our work enhances their interpretability and can accelerate GFlowNet development in practice.

Michael Pühringer, Andreas Hinterreiter, Marc Streit

Classification is one of the most important supervised machine learning tasks. During the training of a classification model, the training instances are fed to the model multiple times (during multiple epochs) in order to iteratively increase the classification performance. The increasing complexity of models has led to a growing demand for model interpretability through visualizations. Existing approaches mostly focus on the visual analysis of the final model performance after training and are often limited to aggregate performance measures. In this paper we introduce InstanceFlow, a novel dual-view visualization tool that allows users to analyze the learning behavior of classifiers over time on the instance-level. A Sankey diagram visualizes the flow of instances throughout epochs, with on-demand detailed glyphs and traces for individual instances. A tabular view allows users to locate interesting instances by ranking and filtering. In this way, InstanceFlow bridges the gap between class-level and instance-level performance evaluation while enabling users to perform a full temporal analysis of the training process.

Andreas Hinterreiter, Marc Streit, Bernhard Kainz

High-dimensional latent representations learned by neural network classifiers are notoriously hard to interpret. Especially in medical applications, model developers and domain experts desire a better understanding of how these latent representations relate to the resulting classification performance. We present Projective Latent Interventions (PLIs), a technique for retraining classifiers by back-propagating manual changes made to low-dimensional embeddings of the latent space. The back-propagation is based on parametric approximations of t-distributed stochastic neighbourhood embeddings. PLIs allow domain experts to control the latent decision space in an intuitive way in order to better match their expectations. For instance, the performance for specific pairs of classes can be enhanced by manually separating the class clusters in the embedding. We evaluate our technique on a real-world scenario in fetal ultrasound imaging.

Andreas Hinterreiter, Christian Steinparz, Moritz Schöfl et al.

In problem-solving, a path towards solutions can be viewed as a sequence of decisions. The decisions, made by humans or computers, describe a trajectory through a high-dimensional representation space of the problem. By means of dimensionality reduction, these trajectories can be visualized in lower-dimensional space. Such embedded trajectories have previously been applied to a wide variety of data, but analysis has focused almost exclusively on the self-similarity of single trajectories. In contrast, we describe patterns emerging from drawing many trajectories -- for different initial conditions, end states, and solution strategies -- in the same embedding space. We argue that general statements about the problem-solving tasks and solving strategies can be made by interpreting these patterns. We explore and characterize such patterns in trajectories resulting from human and machine-made decisions in a variety of application domains: logic puzzles (Rubik's cube), strategy games (chess), and optimization problems (neural network training). We also discuss the importance of suitably chosen representation spaces and similarity metrics for the embedding.

Andreas Hinterreiter, Peter Ruch, Holger Stitz et al.

Classifiers are among the most widely used supervised machine learning algorithms. Many classification models exist, and choosing the right one for a given task is difficult. During model selection and debugging, data scientists need to assess classifiers' performances, evaluate their learning behavior over time, and compare different models. Typically, this analysis is based on single-number performance measures such as accuracy. A more detailed evaluation of classifiers is possible by inspecting class errors. The confusion matrix is an established way for visualizing these class errors, but it was not designed with temporal or comparative analysis in mind. More generally, established performance analysis systems do not allow a combined temporal and comparative analysis of class-level information. To address this issue, we propose ConfusionFlow, an interactive, comparative visualization tool that combines the benefits of class confusion matrices with the visualization of performance characteristics over time. ConfusionFlow is model-agnostic and can be used to compare performances for different model types, model architectures, and/or training and test datasets. We demonstrate the usefulness of ConfusionFlow in a case study on instance selection strategies in active learning. We further assess the scalability of ConfusionFlow and present a use case in the context of neural network pruning.