AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Lifelong Teacher-Student Network Learning

Fei Ye, Adrian G. Borş

TPAMI 2021 · Label-efficient and robust learning › Knowledge distillation · approach: Teacher properties and robustness · 63 citations (OpenAlex)

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 changedWhereMeasuredEffect
add the domain-specific encoder trained with L_delta (ablation)LTS · lifelong MNIST/CIFAR10/Sub-ImageNet/CelebAaverage NLL: 351.09 -> 265.43Improved
add the conditional prior for the domain-specific generative factor (ablation)LTS · lifelong MNIST/CIFAR10/Sub-ImageNet/CelebAaverage NLL: 372.57 -> 265.43Improved
compare LTS to CURL on NLLLTS · four-database lifelongaverage NLL: 315.39 -> 265.43Improved
compare LTS to LGM on NLLLTS · four-database lifelongaverage NLL: 468.28 -> 265.43Improved
LTS vs LGM for lifelong MNIST to SVHN, SVHN accuracyLTS · SVHN M-Saccuracy: 80.77 -> 80.15Hurt
LTS vs LGM, SVHN then MNIST, SVHN accuracyLTS · SVHN S-Maccuracy: 76.76 -> 80.39Improved
LTS vs LGM on MNIST-Fashion F-MLTS · MNIST-Fashion F-Maccuracy: 86.05 -> 89.35Improved
LTS vs LGM on MNIST-Fashion M-FLTS · MNIST-Fashion M-Faccuracy: 91.71 -> 91.49No clear effect
LTS vs LGM on MNIST F-MLTS · MNIST F-Maccuracy: 98.85 -> 98.42Hurt
lifelong semi-supervised LTS vs lifelong LGANLTS · Semi-supervised MNISTclassification error: 4.87 -> 3.18Improved
LTS vs M1+M2 semi-supervised (non-lifelong)LTS · Semi-supervised MNISTclassification error: 2.40 -> 3.18Hurt
LTS vs CURL on SVHN accuracy in three-database lifelongLTS · SVHNaccuracy: 33.53 -> 67.93Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy