Research map / Label-efficient and robust learning / Self-supervised and contrastive
Not All Semantics are Created Equal: Contrastive Self-supervised Learning with Automatic Temperature Individualization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 individualized temperatures learned from the robust global contrastive objective instead of one global temperature | SogCLR-style contrastive learning · CIFAR100-LT | top-1 accuracy: 52.35$\pm$0.08 -> 53.06$\pm$0.12 | Helped |
| Use individualized temperatures learned from the robust global contrastive objective instead of one global temperature | SogCLR-style contrastive learning · Flickr30K image retrieval | Recall@1: 43.32$\pm$0.18 -> 44.36$\pm$0.12 | Helped |
| Use individualized temperatures learned from the robust global contrastive objective instead of one global temperature | SogCLR-style contrastive learning · MSCOCO text retrieval | Recall@1: 30.08$\pm$0.22 -> 32.72$\pm$0.13 | Helped |
| Use individualized temperatures learned from the robust global contrastive objective instead of one global temperature | SogCLR-style contrastive learning · CIFAR10 zero-shot | top-1 accuracy: 61.09$\pm$0.24 -> 58.91$\pm$0.15 | Hurt |
| Vary tau_init across 0.1 to 0.7 (ablation) | iSogCLR · CIFAR10 | top-1 accuracy: 89.00 -> 89.23 | No clear effect |
| Vary rho across 0.1 to 0.4 (ablation) | iSogCLR · CIFAR100 | top-1 accuracy: 63.02 -> 63.82 | No clear effect |
| Set the moving-average parameter beta_0 within 0.7, 0.8, or 0.9 (ablation) | iSogCLR · unimodal image datasets | top-1 accuracy | No clear effect |
Closest papers by meaning
- Model-Aware Contrastive Learning: Towards Escaping the Dilemmas (ICML 2023 · similarity 0.89)
- Multi-modal contrastive learning adapts to intrinsic dimensions of shared latent variables (NeurIPS 2025 · similarity 0.87)
- Self-Supervised Contrastive Learning is Approximately Supervised Contrastive Learning (NeurIPS 2025 · similarity 0.86)
- To Cool or not to Cool? Temperature Network Meets Large Foundation Models via DRO (ICML 2024 · similarity 0.85)
- Improving Diversity in Language Models: When Temperature Fails, Change the Loss (ICML 2025 · similarity 0.85)
- Your contrastive learning problem is secretly a distribution alignment problem (NeurIPS 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.