Research map / Learning theory and algorithms / Clustering
Comparing Apples to Oranges: Learning Similarity Functions for Data Produced by Different Distributions
Leonidas Tsepenekas, Ivan Brugere, Freddy Lecue, Daniele Magazzeni
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 Naive algorithm instead of XGB regressor | similarity learning across groups · Credit Card Default | average relative error (%): 3.121 -> 1.558 | Improved |
| Use Naive algorithm instead of best baseline XGB | similarity learning across groups · Adult | average relative error (%): 2.248 -> 1.319 | Improved |
| Use Naive algorithm instead of best baseline XGB | similarity learning across groups · Give Me Some Credit | average relative error (%): 5.614 -> 1.630 | Improved |
| Use Naive algorithm instead of MLP | similarity learning across groups · Credit Card Default | average absolute error / epsilon: 1.465 -> 0.636 | Improved |
| Use Naive algorithm instead of RF | similarity learning across groups · Give Me Some Credit | average absolute error / epsilon: 7.258 -> 2.644 | Improved |
| Use Naive algorithm instead of RF | similarity learning across groups · Give Me Some Credit | SD of relative error (%): 12.404 -> 4.106 | Improved |
| Use Cluster instead of Naive (fewer queries) | two-phase similarity learner · Credit Card Default | oracle queries: 81.01 | Improved |
| Use Cluster instead of Naive (fewer queries) | two-phase similarity learner · Adult | oracle queries: 80.40 | Improved |
| Use Cluster instead of Naive (fewer queries) | two-phase similarity learner · Give Me Some Credit | oracle queries: 84.87 | Improved |
| Use Cluster instead of Naive | two-phase similarity learner · Credit Card Default | average relative error (%): 1.558 -> 1.593 | No clear effect |
| Use Cluster instead of Naive | two-phase similarity learner · Adult | average relative error (%): 1.319 -> 1.321 | No clear effect |
| Set dilation factor to 2 instead of 8 in Cluster training (ablation) | Cluster · synthetic similarity | accuracy vs queries balance | Improved |
Closest papers by meaning
- Near-Optimal Comparison Based Clustering (NeurIPS 2020 · similarity 0.84)
- Foundations of Comparison-Based Hierarchical Clustering (NeurIPS 2019 · similarity 0.84)
- Twin Learning for Similarity and Clustering: A Unified Kernel Approach (AAAI 2017 · similarity 0.84)
- Average Individual Fairness: Algorithms, Generalization and Experiments (NeurIPS 2019 · similarity 0.84)
- Fairness without Harm: An Influence-Guided Active Sampling Approach (NeurIPS 2024 · similarity 0.83)
- Learning Distributions from Multiple Data Providers (NeurIPS 2026 · similarity 0.82)
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.
