Focus+Context Exploration of Hierarchical Embeddings

Höllt, Thomas; Vilanova, Anna; Pezzotti, Nicola; Lelieveldt, Boudewijn P. F.; Hauser, Helwig

doi:10.1111/cgf.13711

Cited by 11 publications

(14 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In fact, since our approach is agnostic to the type of embedding used, the results of supervised DR techniques can be analyzed with our technique. However, as discussed by H öllt et al [29], additional Focus+Context exploration techniques may be required for an effective analysis of such hierarchical embeddings for large datasets.…”

Section: Hierarchical Embeddingsmentioning

confidence: 99%

Visual Exploration of Relationships and Structure in Low-Dimensional Embeddings

Eckelt

Hinterreiter

Adelberger

et al. 2023

IEEE Trans. Visual. Comput. Graphics

View full text Add to dashboard Cite

In this work, we propose an interactive visual approach for the exploration and formation of structural relationships in embeddings of high-dimensional data. These structural relationships, such as item sequences, associations of items with groups, and hierarchies between groups of items, are defining properties of many real-world datasets. Nevertheless, most existing methods for the visual exploration of embeddings treat these structures as second-class citizens or do not take them into account at all. In our proposed analysis workflow, users explore enriched scatterplots of the embedding, in which relationships between items and/or groups are visually highlighted. The original high-dimensional data for single items, groups of items, or differences between connected items and groups is accessible through additional summary visualizations. We carefully tailored these summary and difference visualizations to the various data types and semantic contexts. During their exploratory analysis, users can externalize their insights by setting up additional groups and relationships between items and/or groups. We demonstrate the utility and potential impact of our approach by means of two use cases and multiple examples from various domains.

show abstract

Section: Hierarchical Embeddingsmentioning

confidence: 99%

Visual Exploration of Relationships and Structure in Low-Dimensional Embeddings

Eckelt

Hinterreiter

Adelberger

et al. 2023

IEEE Trans. Visual. Comput. Graphics

View full text Add to dashboard Cite

show abstract

“…In fact, since our approach is agnostic to the type of embedding used, the results of supervised DR techniques can be analyzed with our technique. However, as discussed by Höllt et al [20], additional Focus+Context exploration techniques may be required for an effective analysis of such hierarchical embeddings for large datasets.…”

Section: Hierarchical Embeddingsmentioning

confidence: 99%

“…Neighborhoods between points within and outside the reprojected part of the embedding may not be interpretable. We are currently investigating how to address this challenge arising from changing only parts of the embedding, potentially based on one of the promising strategies discussed by Höllt et al [20].…”

Section: Focus + Contextmentioning

confidence: 99%

Visual Exploration of Relationships and Structure in Low-Dimensional Embeddings

Eckelt¹,

Hinterreiter²,

Adelberger³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this work, we propose an interactive visual approach for the exploration of structural relationships in embeddings of high-dimensional data. These structural relationships, such as item sequences, associations of items with groups, and hierarchies between groups of items, are defining properties of many real-world datasets. Nevertheless, most existing methods for the visual exploration of embeddings treat these structures as second-class citizens or do not take them into account at all. In our proposed analysis workflow, users explore enriched scatterplots of the embedding, in which relationships between items and/or groups are visually highlighted. The original high-dimensional data for single items, groups of items, or differences between connected items and groups is accessible through additional summary visualizations. We carefully tailored these summary and difference visualizations to the various data types and semantic contexts. During their exploratory analysis, users can externalize their insights by setting up additional groups and relationships between items and/or groups, thereby creating graphs that represent visual data stories. We demonstrate the utility and potential impact of our approach by means of two use cases and multiple examples from various domains.

show abstract

“…requesting focus on these two clusters and generating the projection of Figure 15(B) and (C), the summaries (shown on top) of the two inspected data representatives show that such filter is responsible for detecting the digit borders. Notice that, unlike in Höllt et al's [8] work, users can freely compare the information of different clusters under focus since the tooltip's information is related to data points, not aggregated structures. By aggregating information using the sampling technique and using visual summaries to show information about the clusters, our technique can quickly inform the structures present in the projection due to the facilitated cluster comparison.…”

Section: Use Case -Finding Meaningful Activations Of Cnn Filtersmentioning

confidence: 99%

“…Throughout the exploratory process, users receive visual cues and important structures about the projection. Further, our approach extends the hierarchical exploration concept proposed by Höllt et al [8] even to DR techniques that cannot be extrapolated to hierarchical versions [19]. Since the embedding's hierarchical structure is created based on a sampling selection mechanism [14] that preserves the scatter plots visualiza-tion overview (e.g., class boundaries and outliers), users visualize the main structures with reduced cognitive effort induced by overlapping visual markers.…”

Section: Introductionmentioning

confidence: 99%

ExplorerTree: A Focus+Context Exploration Approach for 2D Embeddings

et al. 2021

View full text Add to dashboard Cite

In exploratory tasks involving high-dimensional datasets, dimensionality reduction (DR) techniques help analysts to discover patterns and other useful information. Although scatter plot representations of DR results allow for cluster identification and similarity analysis, such a visual metaphor presents problems when the number of instances of the dataset increases, resulting in cluttered visualizations. In this work, we propose a scatter plot-based multilevel approach to display DR results and address clutter-related problems when visualizing large datasets, together with the definition of a methodology to use focus+context interaction on non-hierarchical embeddings. The proposed technique, called ExplorerTree, uses a sampling selection technique on scatter plots to reduce visual clutter and guide users through exploratory tasks. We demonstrate ExplorerTree's effectiveness through a use case, where we visually explore activation images of the convolutional layers of a neural network. Finally, we also conducted a user experiment to evaluate ExplorerTree's ability to convey embedding structures using different sampling strategies.

show abstract

Focus+Context Exploration of Hierarchical Embeddings

Cited by 11 publications

References 31 publications

Visual Exploration of Relationships and Structure in Low-Dimensional Embeddings

Visual Exploration of Relationships and Structure in Low-Dimensional Embeddings

Visual Exploration of Relationships and Structure in Low-Dimensional Embeddings

ExplorerTree: A Focus+Context Exploration Approach for 2D Embeddings

Contact Info

Product

Resources

About