AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Co2L: Contrastive Continual Learning

Hyuntak Cha, Jaeho Lee, Jinwoo Shin

ICCV 2021 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Contrastive versus non-contrastive theory · 247 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 IRD (with replay buffer) (ablation)Co2L · Seq-CIFAR-10 Class-ILaccuracy (%): 53.57 -> 65.57Improved
add IRD (without replay buffer) (ablation)Co2L · Seq-CIFAR-10 Class-ILaccuracy (%): 53.25 -> 58.89Improved
add buffered samples to representation learning (with IRD) (ablation)Co2L · Seq-CIFAR-10 Class-ILaccuracy (%): 58.89 -> 65.57Improved
add buffered samples without IRD (ablation)Co2L · Seq-CIFAR-10 Class-ILaccuracy (%): 53.25 -> 53.57No clear effect
use asymmetric SupCon instead of original SupCon (ablation)Co2L · Seq-CIFAR-10 Class-IL (buffer 200)accuracy (%): 60.49 -> 65.57Improved
use asymmetric SupCon instead of original SupCon (ablation)Co2L · Seq-CIFAR-10 Class-IL (buffer 500)accuracy (%): 68.66 -> 74.26Improved
use asymmetric SupCon instead of original SupCon (ablation)Co2L · Seq-Tiny-ImageNet Class-IL (buffer 200)accuracy (%): 13.51 -> 13.88Improved
use asymmetric SupCon instead of original SupCon (ablation)Co2L · Seq-Tiny-ImageNet Class-IL (buffer 500)accuracy (%): 19.68 -> 20.12Improved
use IRD instead of SEED embedding distillation (ablation)Co2L · Seq-CIFAR-10 Class-IL (buffer 200)accuracy (%): 53.42 -> 65.57Improved
use IRD instead of MSE embedding matching (ablation)Co2L · Seq-CIFAR-10 Class-IL (buffer 500)accuracy (%): 62.83 -> 74.26Improved
use IRD instead of MSE projection matching (ablation)Co2L · Seq-CIFAR-10 Class-IL (buffer 500)accuracy (%): 57.47 -> 74.26Improved
use IRD instead of MSE projection matching (ablation)Co2L · Seq-Tiny-ImageNet Task-IL (buffer 200)accuracy (%): 34.38 -> 42.37Improved

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