Research map / Label-efficient and robust learning / Self-supervised and contrastive
Self-Supervised Generalisation with Meta Auxiliary Learning
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 |
|---|---|---|---|
| add MAXL auxiliary-label learning (ψ[i]=3) | 4-layer ConvNet · MNIST | accuracy: 99.57±0.02 -> 99.71 ± 0.02 | Helped |
| add MAXL auxiliary-label learning (ψ[i]=5) | 4-layer ConvNet · SVHN | accuracy: 94.05 ± 0.07 -> 94.59 ± 0.12 | Helped |
| add MAXL auxiliary-label learning (ψ[i]=5) | VGG-16 · CIFAR-10 | accuracy: 92.77 ± 0.13 -> 93.49 ± 0.05 | Helped |
| add MAXL auxiliary-label learning (ψ[i]=5) | VGG-16 · ImageNet | accuracy: 46.67 ± 0.12 -> 47.02 ± 0.11 | Helped |
| add MAXL auxiliary-label learning (ψ[i]=5) | ResNet-32 · CINIC-10 | accuracy: 85.12 ± 0.08 -> 85.83 ± 0.08 | Helped |
| add MAXL auxiliary-label learning (ψ[i]=3) | ResNet-32 · UCF-101 | accuracy: 53.15 ± 0.12 -> 55.39 ± 0.16 | Helped |
| add entropy regularization to MAXL (ablation) | MAXL · CIFAR-100, 3 primary classes and 10 auxiliary classes | accuracy: 90.26 -> 90.50 | Helped |
| add entropy regularization to MAXL (ablation) | MAXL · CIFAR-100, 3 primary classes and 20 auxiliary classes | accuracy: 90.39 -> 90.65 | Helped |
| add entropy regularization to MAXL (ablation) | MAXL · CIFAR-100, 3 primary classes and 100 auxiliary classes | accuracy: 90.22 -> 90.66 | Helped |
| add entropy regularization to MAXL (ablation) | MAXL · CIFAR-100, 10 primary classes and 20 auxiliary classes | accuracy: 77.73 -> 78.40 | Helped |
| add entropy regularization to MAXL (ablation) | MAXL · CIFAR-100, 10 primary classes and 100 auxiliary classes | accuracy: 78.20 -> 78.46 | Helped |
| add entropy regularization to MAXL (ablation) | MAXL · CIFAR-100, 20 primary classes and 100 auxiliary classes | accuracy: 73.97 -> 74.27 | Helped |
Closest papers by meaning
- Bridging Multi-Task Learning and Meta-Learning: Towards Efficient Training and Effective Adaptation (ICML 2021 · similarity 0.89)
- Meta-Learning without Memorization (ICLR 2020 · similarity 0.88)
- How to Distribute Data across Tasks for Meta-Learning? (AAAI 2022 · similarity 0.87)
- Aux-NAS: Exploiting Auxiliary Labels with Negligibly Extra Inference Cost (ICLR 2024 · similarity 0.87)
- Improving Generalization via Meta-Learning on Hard Samples (CVPR 2024 · similarity 0.86)
- Unsupervised Meta-Learning via In-Context Learning (ICLR 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.