AI Paper Atlas

Research map / Continual learning / Online and replay-based

Rethinking Gradient Projection Continual Learning: Stability / Plasticity Feature Space Decoupling

Zhen Zhao, Zhizhong Zhang, Xin Tan, Jun Liu, Yanyun Qu, Yuan Xie, Lizhuang Ma

CVPR 2023 · Continual learning › Online and replay-based · approach: Unified and unsupervised continual frameworks · 24 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
Apply Space Decoupling to TRGPTRGP · 20-split-MiniImageNetACC (%): +1.98%Improved
Apply Space Decoupling to TRGPTRGP · 20-split-CIFAR-100ACC (%): +3.16%Improved
Apply Space Decoupling to TRGPTRGP · 10-split-CIFAR-100ACC (%): +1.04%Improved
Add I/R Approximation (ablation)GPM · 20-split-CIFAR-100ACC (%): 77.53±0.83 -> 79.99±0.48Improved
Add I/R Approximation (ablation)TRGP · 20-split-CIFAR-100ACC (%): 80.68±0.7 -> 83.21±0.36Improved
Add I/R Projection (ablation)GPM · 20-split-CIFAR-100ACC (%): 77.53±0.83 -> 78.41±1.3Improved
Add I/R Projection (ablation)GPM · 20-split-CIFAR-100BWT (%): -0.97±0.59 -> -1.44±0.65Hurt
Add I/R Projection (ablation)TRGP · 20-split-CIFAR-100ACC (%): 80.68±0.7 -> 81.15±1.22Improved
Add I/R Projection (ablation)TRGP · 20-split-CIFAR-100BWT (%): -0.87±0.46 -> -1.24±0.71Hurt
Add both I/R Approximation and I/R Projection (ablation)GPM · 20-split-CIFAR-100ACC (%): 77.53±0.83 -> 80.71±0.82Improved
Reduce the feature-space dimension using I/R Approximation (ablation)GPM · 10-split-CIFAR-100, 20-split-CIFAR-100, 20-split-MiniImageNetACCImproved
Construct stability and residual subspaces by manually selecting top singular-value bases instead of intersections (ablation)GPM+SD and TRGP+SD · 10-split-CIFAR-100, 20-split-CIFAR-100ACC: 77.1; 80.44; 78.32; 80.75; 83.92; 80.64Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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