Research map / Label-efficient and robust learning / Self-supervised and contrastive
Contrasting Quadratic Assignments for Set-Based Representation Learning
Artem Moskalev, Ivan Sosnovik, Volker Fischer, Arnold Smeulders
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 |
|---|---|---|---|
| add QARe regularization | Triplet · CUHK-03 | matching accuracy (%): 54.85 -> 58.48 | Improved |
| add QARe regularization | InfoNCE · CUHK-03 | matching accuracy (%): 57.87 -> 61.96 | Improved |
| add QARe regularization | NTLogistic · CUHK-03 | matching accuracy (%): 40.30 -> 43.48 | Improved |
| add QARe regularization | SparseCLR · CUHK-03 | matching accuracy (%): 53.03 -> 54.84 | Improved |
| add QARe regularization | SimCLR Conv-4 · CIFAR-10 | accuracy (%): 59.64 -> 61.54 | Improved |
| add QARe regularization | SimCLR ResNet-32 · tiny-ImageNet | accuracy (%): 23.44 -> 25.05 | Improved |
| add QARe regularization | SimCLR Conv-4 · tiny-ImageNet | accuracy (%): 21.15 -> 21.33 | No clear effect |
| add QARe regularization | SimCLR ResNet-32 · CIFAR-10 | accuracy (%): 76.45 -> 76.85 | No clear effect |
| add QARe regularization | SparseCLR ResNet-32 · tiny-ImageNet | accuracy (%): 25.87 -> 27.03 | Improved |
| add QARe regularization | SparseCLR Conv-4 · CIFAR-100 | accuracy (%): 32.04 -> 33.05 | Improved |
| replace log-sum-exp (InfoNCE-style) smoothing with sparsemax (SparseCLR) | SimCLR ResNet-32 · CIFAR-10 | accuracy (%): 76.45 -> 70.37 | Hurt |
| replace log-sum-exp (InfoNCE-style) smoothing with sparsemax (SparseCLR) | SimCLR ResNet-32 · tiny-ImageNet | accuracy (%): 23.44 -> 25.87 | Improved |
Closest papers by meaning
- On the Similarities of Embeddings in Contrastive Learning (ICML 2025 · similarity 0.86)
- Embedding Dimension of Contrastive Learning and $k$-Nearest Neighbors (NeurIPS 2024 · similarity 0.85)
- $\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs (ICLR 2025 · similarity 0.85)
- REBAR: Retrieval-Based Reconstruction for Time-series Contrastive Learning (ICLR 2024 · similarity 0.85)
- Weakly-Supervised Contrastive Learning for Imprecise Class Labels (ICML 2025 · similarity 0.85)
- Perfect Alignment May be Poisonous to Graph Contrastive Learning (ICML 2024 · similarity 0.84)
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.
