AI Paper Atlas

Research map / Continual learning / Online and replay-based

Residual Continual Learning

Janghyeon Lee, Donggyu Joo, Hyeong Gwon Hong, Junmo Kim

AAAI 2020 · Continual learning › Online and replay-based · approach: Theoretical analysis of forgetting · 20 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 ResCL's layer-wise residual reparameterization and decay loss instead of LwFpre-activation ResNet-32 · CIFAR-10 → CIFAR-100maximum achievable average accuracy (%): 77.67±0.10 -> 78.80±0.17Improved
Use ResCL's layer-wise residual reparameterization and decay loss instead of Mean-IMMpre-activation ResNet-32 · CIFAR-100 → CIFAR-10maximum achievable average accuracy (%): 76.41±0.19 -> 77.13±0.12Improved
Use ResCL's layer-wise residual reparameterization and decay loss instead of Mean-IMMpre-activation ResNet-32 · CIFAR-10 → SVHNmaximum achievable average accuracy (%): 79.91±1.47 -> 89.49±0.32Improved
Use ResCL instead of LwFpre-activation ResNet · CIFAR-10 → CIFAR-100 → SVHNmaximum achievable average accuracy (%): 58.44 -> 78.53Improved
Use ResCL instead of Mean-IMMAlexNet · ImageNet → CUBmaximum achievable average accuracy (%): 52.12 -> 53.51Improved
Use ResCL instead of LwFVGG · ImageNet → CUBmaximum achievable average accuracy (%): 68.82 -> 68.95Improved
Use ResCL instead of LwF at the required target accuracypre-activation ResNet-32 · CIFAR-10 → CIFAR-100source accuracy (%): 89.59 -> 90.65Improved
Use ResCL instead of LwF at the required target accuracypre-activation ResNet-32 · CIFAR-100 → CIFAR-10source accuracy (%): 63.91 -> 68.13Improved
Use ResCL instead of LwF at the required target accuracypre-activation ResNet-32 · CIFAR-10 → SVHNsource accuracy (%): 38.08 -> 76.83Improved
Increase the decay-loss multiplier λResCL on pre-activation ResNet-32 · CIFAR-10 → SVHNsource accuracyImproved
Increase the decay-loss multiplier λResCL on pre-activation ResNet-32 · CIFAR-10 → SVHNtarget accuracyHurt

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