List of publications and preprints by Tom Hanika
2024
- 1.Hirth, J., Horn, V., Stumme, G., Hanika, T.: Ordinal motifs in lattices. Information Sciences. 659, 120009 (2024). https://doi.org/https://doi.org/10.1016/j.ins.2023.120009.
@article{HIRTH2024120009,
author = {Hirth, Johannes and Horn, Viktoria and Stumme, Gerd and Hanika, Tom},
journal = {Information Sciences},
keywords = {itegpub},
pages = 120009,
title = {Ordinal motifs in lattices},
volume = 659,
year = 2024
}%0 Journal Article
%1 HIRTH2024120009
%A Hirth, Johannes
%A Horn, Viktoria
%A Stumme, Gerd
%A Hanika, Tom
%D 2024
%J Information Sciences
%P 120009
%R https://doi.org/10.1016/j.ins.2023.120009
%T Ordinal motifs in lattices
%U https://www.sciencedirect.com/science/article/pii/S0020025523015943
%V 659 - 1.Hirth, J., Hanika, T.: The Geometric Structure of Topic Models, (2024).Topic models are a popular tool for clustering and analyzing textual data. They allow texts to be classified on the basis of their affiliation to the previously calculated topics. Despite their widespread use in research and application, an in-depth analysis of topic models is still an open research topic. State-of-the-art methods for interpreting topic models are based on simple visualizations, such as similarity matrices, top-term lists or embeddings, which are limited to a maximum of three dimensions. In this paper, we propose an incidence-geometric method for deriving an ordinal structure from flat topic models, such as non-negative matrix factorization. These enable the analysis of the topic model in a higher (order) dimension and the possibility of extracting conceptual relationships between several topics at once. Due to the use of conceptual scaling, our approach does not introduce any artificial topical relationships, such as artifacts of feature compression. Based on our findings, we present a new visualization paradigm for concept hierarchies based on ordinal motifs. These allow for a top-down view on topic spaces. We introduce and demonstrate the applicability of our approach based on a topic model derived from a corpus of scientific papers taken from 32 top machine learning venues.
@preprint{hirth2024geometric,
abstract = {Topic models are a popular tool for clustering and analyzing textual data. They allow texts to be classified on the basis of their affiliation to the previously calculated topics. Despite their widespread use in research and application, an in-depth analysis of topic models is still an open research topic. State-of-the-art methods for interpreting topic models are based on simple visualizations, such as similarity matrices, top-term lists or embeddings, which are limited to a maximum of three dimensions. In this paper, we propose an incidence-geometric method for deriving an ordinal structure from flat topic models, such as non-negative matrix factorization. These enable the analysis of the topic model in a higher (order) dimension and the possibility of extracting conceptual relationships between several topics at once. Due to the use of conceptual scaling, our approach does not introduce any artificial topical relationships, such as artifacts of feature compression. Based on our findings, we present a new visualization paradigm for concept hierarchies based on ordinal motifs. These allow for a top-down view on topic spaces. We introduce and demonstrate the applicability of our approach based on a topic model derived from a corpus of scientific papers taken from 32 top machine learning venues.},
author = {Hirth, Johannes and Hanika, Tom},
keywords = {kde},
title = {The Geometric Structure of Topic Models},
year = 2024
}%0 Generic
%1 hirth2024geometric
%A Hirth, Johannes
%A Hanika, Tom
%D 2024
%T The Geometric Structure of Topic Models
%X Topic models are a popular tool for clustering and analyzing textual data. They allow texts to be classified on the basis of their affiliation to the previously calculated topics. Despite their widespread use in research and application, an in-depth analysis of topic models is still an open research topic. State-of-the-art methods for interpreting topic models are based on simple visualizations, such as similarity matrices, top-term lists or embeddings, which are limited to a maximum of three dimensions. In this paper, we propose an incidence-geometric method for deriving an ordinal structure from flat topic models, such as non-negative matrix factorization. These enable the analysis of the topic model in a higher (order) dimension and the possibility of extracting conceptual relationships between several topics at once. Due to the use of conceptual scaling, our approach does not introduce any artificial topical relationships, such as artifacts of feature compression. Based on our findings, we present a new visualization paradigm for concept hierarchies based on ordinal motifs. These allow for a top-down view on topic spaces. We introduce and demonstrate the applicability of our approach based on a topic model derived from a corpus of scientific papers taken from 32 top machine learning venues. - 1.Hille, T., Stubbemann, M., Hanika, T.: Reproducibility and Geometric Intrinsic Dimensionality: An Investigation on Graph Neural Network Research, (2024).
@preprint{hille2024reproducibility,
author = {Hille, Tobias and Stubbemann, Maximilian and Hanika, Tom},
keywords = {kde},
title = {Reproducibility and Geometric Intrinsic Dimensionality: An Investigation on Graph Neural Network Research},
year = 2024
}%0 Generic
%1 hille2024reproducibility
%A Hille, Tobias
%A Stubbemann, Maximilian
%A Hanika, Tom
%D 2024
%T Reproducibility and Geometric Intrinsic Dimensionality: An Investigation on Graph Neural Network Research
2023
- 1.Stubbemann, M., Hanika, T., Schneider, F.M.: Intrinsic Dimension for Large-Scale Geometric Learning. Transactions on Machine Learning Research. (2023).The concept of dimension is essential to grasp the complexity of data. A naive approach to determine the dimension of a dataset is based on the number of attributes. More sophisticated methods derive a notion of intrinsic dimension (ID) that employs more complex feature functions, e.g., distances between data points. Yet, many of these approaches are based on empirical observations, cannot cope with the geometric character of contemporary datasets, and do lack an axiomatic foundation. A different approach was proposed by V. Pestov, who links the intrinsic dimension axiomatically to the mathematical concentration of measure phenomenon. First methods to compute this and related notions for ID were computationally intractable for large-scale real-world datasets. In the present work, we derive a computationally feasible method for determining said axiomatic ID functions. Moreover, we demonstrate how the geometric properties of complex data are accounted for in our modeling. In particular, we propose a principle way to incorporate neighborhood information, as in graph data, into the ID. This allows for new insights into common graph learning procedures, which we illustrate by experiments on the Open Graph Benchmark.
@article{stubbemann2022intrinsic,
abstract = {The concept of dimension is essential to grasp the complexity of data. A naive approach to determine the dimension of a dataset is based on the number of attributes. More sophisticated methods derive a notion of intrinsic dimension (ID) that employs more complex feature functions, e.g., distances between data points. Yet, many of these approaches are based on empirical observations, cannot cope with the geometric character of contemporary datasets, and do lack an axiomatic foundation. A different approach was proposed by V. Pestov, who links the intrinsic dimension axiomatically to the mathematical concentration of measure phenomenon. First methods to compute this and related notions for ID were computationally intractable for large-scale real-world datasets. In the present work, we derive a computationally feasible method for determining said axiomatic ID functions. Moreover, we demonstrate how the geometric properties of complex data are accounted for in our modeling. In particular, we propose a principle way to incorporate neighborhood information, as in graph data, into the ID. This allows for new insights into common graph learning procedures, which we illustrate by experiments on the Open Graph Benchmark.},
author = {Stubbemann, Maximilian and Hanika, Tom and Schneider, Friedrich Martin},
journal = {Transactions on Machine Learning Research},
keywords = {itegpub},
title = {Intrinsic Dimension for Large-Scale Geometric Learning},
year = 2023
}%0 Journal Article
%1 stubbemann2022intrinsic
%A Stubbemann, Maximilian
%A Hanika, Tom
%A Schneider, Friedrich Martin
%D 2023
%J Transactions on Machine Learning Research
%T Intrinsic Dimension for Large-Scale Geometric Learning
%U https://openreview.net/forum?id=85BfDdYMBY
%X The concept of dimension is essential to grasp the complexity of data. A naive approach to determine the dimension of a dataset is based on the number of attributes. More sophisticated methods derive a notion of intrinsic dimension (ID) that employs more complex feature functions, e.g., distances between data points. Yet, many of these approaches are based on empirical observations, cannot cope with the geometric character of contemporary datasets, and do lack an axiomatic foundation. A different approach was proposed by V. Pestov, who links the intrinsic dimension axiomatically to the mathematical concentration of measure phenomenon. First methods to compute this and related notions for ID were computationally intractable for large-scale real-world datasets. In the present work, we derive a computationally feasible method for determining said axiomatic ID functions. Moreover, we demonstrate how the geometric properties of complex data are accounted for in our modeling. In particular, we propose a principle way to incorporate neighborhood information, as in graph data, into the ID. This allows for new insights into common graph learning procedures, which we illustrate by experiments on the Open Graph Benchmark. - 1.Dürrschnabel, D., Hanika, T., Stumme, G.: Drawing Order Diagrams Through Two-Dimension Extension. Journal of Graph Algorithms and Applications. 27, 783–802 (2023). https://doi.org/10.7155/jgaa.00645.
@article{drrschnabel2023drawing,
author = {Dürrschnabel, Dominik and Hanika, Tom and Stumme, Gerd},
journal = {Journal of Graph Algorithms and Applications},
keywords = {itegpub},
number = 9,
pages = {783–802},
publisher = {Journal of Graph Algorithms and Applications},
title = {Drawing Order Diagrams Through Two-Dimension Extension},
volume = 27,
year = 2023
}%0 Journal Article
%1 drrschnabel2023drawing
%A Dürrschnabel, Dominik
%A Hanika, Tom
%A Stumme, Gerd
%D 2023
%I Journal of Graph Algorithms and Applications
%J Journal of Graph Algorithms and Applications
%N 9
%P 783–802
%R 10.7155/jgaa.00645
%T Drawing Order Diagrams Through Two-Dimension Extension
%U http://dx.doi.org/10.7155/jgaa.00645
%V 27