Research map / Label-efficient and robust learning / Knowledge distillation
Lifelong Teacher-Student Network Learning
Fei Ye, Adrian G. Borş
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 the domain-specific encoder trained with L_delta (ablation) | LTS · lifelong MNIST/CIFAR10/Sub-ImageNet/CelebA | average NLL: 351.09 -> 265.43 | Improved |
| add the conditional prior for the domain-specific generative factor (ablation) | LTS · lifelong MNIST/CIFAR10/Sub-ImageNet/CelebA | average NLL: 372.57 -> 265.43 | Improved |
| compare LTS to CURL on NLL | LTS · four-database lifelong | average NLL: 315.39 -> 265.43 | Improved |
| compare LTS to LGM on NLL | LTS · four-database lifelong | average NLL: 468.28 -> 265.43 | Improved |
| LTS vs LGM for lifelong MNIST to SVHN, SVHN accuracy | LTS · SVHN M-S | accuracy: 80.77 -> 80.15 | Hurt |
| LTS vs LGM, SVHN then MNIST, SVHN accuracy | LTS · SVHN S-M | accuracy: 76.76 -> 80.39 | Improved |
| LTS vs LGM on MNIST-Fashion F-M | LTS · MNIST-Fashion F-M | accuracy: 86.05 -> 89.35 | Improved |
| LTS vs LGM on MNIST-Fashion M-F | LTS · MNIST-Fashion M-F | accuracy: 91.71 -> 91.49 | No clear effect |
| LTS vs LGM on MNIST F-M | LTS · MNIST F-M | accuracy: 98.85 -> 98.42 | Hurt |
| lifelong semi-supervised LTS vs lifelong LGAN | LTS · Semi-supervised MNIST | classification error: 4.87 -> 3.18 | Improved |
| LTS vs M1+M2 semi-supervised (non-lifelong) | LTS · Semi-supervised MNIST | classification error: 2.40 -> 3.18 | Hurt |
| LTS vs CURL on SVHN accuracy in three-database lifelong | LTS · SVHN | accuracy: 33.53 -> 67.93 | Improved |
Closest papers by meaning
- Dynamic Self-Supervised Teacher-Student Network Learning (TPAMI 2022 · similarity 0.90)
- Lifelong GAN: Continual Learning for Conditional Image Generation (ICCV 2019 · similarity 0.87)
- GAN Memory with No Forgetting (NeurIPS 2020 · similarity 0.87)
- Adversarial Learning of Portable Student Networks (AAAI 2018 · similarity 0.87)
- Hybrid Data-Free Knowledge Distillation (AAAI 2025 · similarity 0.86)
- Recall-Oriented Continual Learning with Generative Adversarial Meta-Model (AAAI 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.
