AI Paper Atlas

Research map / Continual learning / Domain and prompt-based

Lifelong Domain Adaptation via Consolidated Internal Distribution

Mohammad Rostami

NeurIPS 2021 · Continual learning › Domain and prompt-based · approach: Domain-incremental learning with memory · 23 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
use LDAuCID adaptation instead of no adaptation (source only) (ablation)VGG16 / ResNet-50 models · SVHN->MNISTaccuracy: 67.3±2.6 -> 91.4 ± 2.2Improved
use LDAuCID adaptation instead of no adaptation (source only) (ablation)ResNet-50 · Office-Home averageaccuracy: 46.1 -> 59.4Improved
use LDAuCID without source data instead of joint-training UDA methods (CDAN)LDAuCID · Office-Home averageaccuracy: 63.9 -> 59.4Hurt
use LDAuCID instead of ETDLDAuCID · SVHN->MNISTaccuracy: 97.9± 0.4 -> 91.4 ± 2.2Hurt
use LDAuCID instead of ETDLDAuCID · ImageCLEF-DA averageaccuracy: 89.7 -> 95.8Improved
increase memory buffer size from 50 to 100 samples (ablation)LDAuCID · S->M->U, USPSaccuracy: 91.9 -> 94.6Improved
make training data imbalanced across digit classes (ablation)LDAuCID · S->M->U, USPSaccuracy: 93.1 on imbalanced data (balanced value not tabulated)Hurt
vary trade-off parameter lambda (ablation)LDAuCID · SVHN->MNISTaccuracyNo clear effect
raise pseudo-sample confidence threshold tau toward 1 (ablation)LDAuCID · SVHN->MNISTaccuracyImproved

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