Research map / Graph learning / Graph self-supervised learning
Learning Representations of Persistence Barcodes
Christoph D. Hofer, Roland Kwitt, Marc Niethammer
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| learn the structure element parameters instead of only k-means++ initialization (frozen) (ablation) | persistence barcode input layer · MPEG-7 | accuracy (%): 92.2 -> 92.7 | Improved |
| learn the structure element parameters instead of frozen k-means++ initialization (ablation) | persistence barcode input layer · Animal | accuracy (%): 72.9 -> 74.6 | Improved |
| learn the structure element parameters instead of frozen k-means++ initialization (ablation) | persistence barcode input layer · SHREC14 (Real) | accuracy (%): 70.5 -> 72.2 | Improved |
| k-means++ initialization instead of random initialization of structure elements (ablation) | input layer · MPEG-7 | accuracy (%) | Improved |
| learning the parametrization instead of freezing it (ablation) | input layer · MPEG-7 (varying N) | accuracy (%) | Improved |
| persistence images fed to a CNN instead of a linear SVM | persistence images · MPEG-7 | accuracy (%): 85.1 -> 92.3 | Improved |
| persistence images fed to a CNN instead of a linear SVM | persistence images · Animal | accuracy (%): 61.1 -> 69.4 | Improved |
| proposed learned layer instead of PSS kernel | Ours (learned) · MPEG-7 | accuracy (%): 94.5 vs 92.7 | Hurt |
| proposed learned layer instead of PSS kernel | Ours (learned) · Animal | accuracy (%): 73.6 -> 74.6 | Improved |
| proposed learned layer instead of persistence images with NN | Ours (learned) · SHREC14 (Real) | accuracy (%): 69.0 -> 72.2 | Improved |
| adding more convolution layers to persistence-image CNN (ablation) | persistence images NN · benchmarks | accuracy (%) | No clear effect |
| different initializations of r within a reasonable range (ablation) | Ours · benchmarks | accuracy (%) | No clear effect |
Closest papers by meaning
- Persistence Paths and Signature Features in Topological Data Analysis (TPAMI 2018 · similarity 0.92)
- A Survey of Vectorization Methods in Topological Data Analysis (TPAMI 2023 · similarity 0.91)
- Graphcode: Learning from multiparameter persistent homology using graph neural networks (NeurIPS 2024 · similarity 0.90)
- Persistence Images: A Stable Vector Representation of Persistent Homology (JMLR 2017 · similarity 0.89)
- Learning metrics for persistence-based summaries and applications for graph classification (NeurIPS 2019 · similarity 0.89)
- Differentiability and Optimization of Multiparameter Persistent Homology (ICML 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
