Research map / Learning theory and algorithms / Prediction and decision losses
Relevance-Based Embeddings: Lightweight Candidate Retrieval via Heavy-Ranker Calls
Kirill Sergeevich Shevkunov, Andrey Ploskonosov, Liudmila Prokhorenkova
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 |
|---|---|---|---|
| select support items by l2-greedy instead of random (ablation) | AnnCUR · Yugioh | HitRate(100): 0.4724 -> 0.5618 | Improved |
| select support items by l2-greedy instead of random (ablation) | AnnCUR · Dr.Who | HitRate(100): 0.1919 -> 0.2960 | Improved |
| select support items by l2-greedy instead of random (ablation) | AnnCUR · QA.Small | HitRate(100): 0.5828 -> 0.6100 | Improved |
| select support items by l2-greedy instead of random (ablation) | AnnCUR · RecMusic | HitRate(100): 0.1478 -> 0.1478 | No clear effect |
| select popular items as support items instead of random (ablation) | AnnCUR · Yugioh | HitRate(100): 0.4724 -> 0.2429 | Hurt |
| select popular items as support items instead of random (ablation) | AnnCUR · RecGames2 | HitRate(100): 0.6697 -> 0.7623 | Improved |
| select KMeans cluster centers as support items instead of random (ablation) | AnnCUR · StarTrek | HitRate(100): 0.2287 -> 0.3226 | Improved |
| select KMeans cluster centers as support items instead of random (ablation) | AnnCUR · QA.Small | HitRate(100): 0.5828 -> 0.5578 | Hurt |
| add trainable relevance mappings (RBE) to KMeans support items | AnnCUR+KMeans · RecMusic | HitRate(100): 0.1661 -> 0.3729 | Improved |
| add trainable relevance mappings (RBE) to l2-greedy support items | AnnCUR+l2-greedy · QA | HitRate(100): 0.5700 -> 0.6022 | Improved |
| add trainable relevance mappings (RBE) to l2-greedy support items | AnnCUR+l2-greedy · Military | HitRate(100): 0.3357 -> 0.3349 | No clear effect |
| use RBE instead of dual encoder | RecGames1 · RecGames1 (X=100) | HitRate(X,X): 0.7048 -> 0.6682 | Hurt |
Closest papers by meaning
- NUDGE: Lightweight Non-Parametric Fine-Tuning of Embeddings for Retrieval (ICLR 2025 · similarity 0.88)
- Adaptive Retrieval and Scalable Indexing for k-NN Search with Cross-Encoders (ICLR 2024 · similarity 0.86)
- MUVERA: Multi-Vector Retrieval via Fixed Dimensional Encoding (NeurIPS 2024 · similarity 0.86)
- USTAD: Unified Single-model Training Achieving Diverse Scores for Information Retrieval (ICML 2024 · similarity 0.86)
- Exploiting Distribution Constraints for Scalable and Efficient Image Retrieval (ICLR 2025 · similarity 0.85)
- Evaluating Transferability in Retrieval Tasks: An Approach Using MMD and Kernel Methods (CVPR 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.
