Research map / Optimization / Optimizers
Ultrametric Fitting by Gradient Descent
Giovanni Chierchia, Benjamin Perret
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 |
|---|---|---|---|
| Use gradient descent with the min-max ultrametric operator to fit the dissimilarity graph | Ultrametric fitting · Planar superpixel graphs | mean square error | Improved |
| Use Closest+Size or Dasgupta instead of Ward | Hierarchical clustering · Five LIBSVM datasets | classification accuracy | No clear effect |
| Use Closest+Triplet with labeled triplets | Semi-supervised clustering · Five LIBSVM datasets | classification accuracy | No clear effect |
| Use semi-supervised SVM instead of spectral clustering | Semi-supervised clustering · Five LIBSVM datasets | classification accuracy | Improved |
| Use the Dasgupta relaxation instead of Closest+Size | Hierarchical clustering · Digit1 and Heart | classification accuracy | Hurt |
| Use average linkage or closest ultrametric instead of Ward | Hierarchical clustering · Five LIBSVM datasets | classification accuracy | Hurt |
| Add cluster-size regularization to closest-ultrametric fitting (ablation) | Closest ultrametric · Illustrative hierarchical-clustering examples | small clusters at large scales | Improved |
| Add triplet regularization to closest-ultrametric fitting (ablation) | Closest ultrametric · Semi-supervised clustering | classification accuracy | Improved |
| Replace hard node cardinality with a differentiable soft-cardinal measure (ablation) | Dasgupta cost · Hierarchical clustering | classification accuracy | No clear effect |
Closest papers by meaning
- Ultrametric Cluster Hierarchies: I Want ‘em All! (NeurIPS 2025 · similarity 0.88)
- Hierarchical Clustering with Structural Constraints (ICML 2018 · similarity 0.88)
- New Bounds on the Cohesion of Complete-link and Other Linkage Methods for Agglomerative Clustering (ICML 2024 · similarity 0.86)
- A (1+?)-Approximation for Ultrametric Embedding in Subquadratic Time (AAAI 2025 · similarity 0.86)
- On the cohesion and separability of average-link for hierarchical agglomerative clustering (NeurIPS 2024 · similarity 0.86)
- Expected Probabilistic Hierarchies (NeurIPS 2024 · similarity 0.85)
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.
