Research map / Label-efficient and robust learning / Self-supervised and contrastive
Scarf: Self-Supervised Contrastive Learning using Random Feature Corruption
Dara Bahri, Heinrich Jiang, Yi Tay, Donald Metzler
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 SCARF contrastive pre-training | Control neural network · OpenML-CC18, 100% labeled | relative gain in accuracy: 2.352 | Improved |
| Add SCARF contrastive pre-training | Control neural network · OpenML-CC18, 25% labeled | relative gain in accuracy: 3.692 | Improved |
| Add SCARF contrastive pre-training | Control neural network · OpenML-CC18, 30% label noise | relative gain in accuracy: 2.261 | Improved |
| Add SCARF pre-training | Deep k-NN · OpenML-CC18, 30% label noise | relative gain in accuracy: 2.001 | Improved |
| Replace empirical marginal feature sampling with no corruption (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
| Replace empirical marginal sampling with mean replacement (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
| Replace marginal sampling with additive Gaussian noise (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
| Replace independent marginal draws with joint training-data sampling (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
| Replace marginal sampling with missing-feature values (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
| Replace marginal sampling with feature dropout (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
| Corrupt both views instead of only one (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
| Use the same corrupted feature indices for every example in a batch (ablation) | SCARF pre-training · OpenML-CC18 classification | accuracy | Hurt |
Later papers that compared with this method
1 later accepted paper reports SCARF as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- TransTab: Learning Transferable Tabular Transformers Across Tables (NeurIPS 2022 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Self-Supervised Contrastive Learning is Approximately Supervised Contrastive Learning (NeurIPS 2025 · similarity 0.88)
- Self-Contrastive Learning: Single-Viewed Supervised Contrastive Framework Using Sub-network (AAAI 2023 · similarity 0.88)
- Self-Damaging Contrastive Learning (ICML 2021 · similarity 0.87)
- Data-Efficient Contrastive Self-supervised Learning: Most Beneficial Examples for Supervised Learning Contribute the Least (ICML 2023 · similarity 0.87)
- Divide and Contrast: Self-Supervised Learning From Uncurated Data (ICCV 2021 · similarity 0.87)
- An Augmentation-Aware Theory for Self-Supervised Contrastive Learning (ICML 2025 · 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.
