Research map / Learning theory and algorithms / Prediction and decision losses
Insights into Ordinal Embedding Algorithms: A Systematic Evaluation
Leena Chennuru Vankadara, Michael Lohaus, Siavash Haghiri, Faiz Ul Wahab, Ulrike von Luxburg
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 t-STE instead of STE | STE · MNIST | test triplet error: 0.11 -> 0.2 | Hurt |
| use t-STE instead of STE | STE · COVTYPE | kNN error: 0.5 -> 0.53 | Hurt |
| optimize over the embedding matrix (SOE) instead of Gram matrix (GNMDS) | GNMDS · USPS | test triplet error: 0.35 -> 0.02 | Improved |
| optimize over the embedding matrix (SOE) instead of Gram matrix (CKL) | CKL · USPS | kNN error: 0.45 -> 0.04 | Improved |
| use FORTE instead of other Gram-matrix methods | GNMDS · USPS | test triplet error: 0.35 -> 0.06 | Improved |
| use SOE instead of landmark method LLOE | LLOE · MNIST | test triplet error: 0.31 -> 0.03 | Improved |
| use SOE instead of neural method OENN | OENN · MNIST | test triplet error: 0.13 -> 0.03 | Improved |
| use SOE instead of tSTE | tSTE · COVTYPE | test triplet error: 0.24 -> 0.02 | Improved |
| use SOE instead of OENN on COVTYPE | OENN · COVTYPE | kNN error: 0.29 vs 0.13 | Hurt |
| choose triplets at random instead of the landmark selection procedure for SOE | SOE · general | convergence time | Improved |
| increase noise level in triplets | all OE methods · Aggregation | reconstruction quality | Hurt |
| increase embedding dimension | SOE, STE, tSTE, OENN · USPS | kNN and Procrustes error | Improved |
Closest papers by meaning
- Scaling Up Ordinal Embedding: A Landmark Approach (ICML 2019 · similarity 0.93)
- Finite Sample Prediction and Recovery Bounds for Ordinal Embedding (NeurIPS 2016 · similarity 0.92)
- On the rankability of visual embeddings (NeurIPS 2025 · similarity 0.86)
- Euclidean Embedding of Data Using Local Distances (NeurIPS 2026 · similarity 0.85)
- Weighted Embeddings for Low-Dimensional Graph Representation (AAAI 2025 · similarity 0.85)
- When is an Embedding Model More Promising than Another? (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.
