AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Imbalanced Continual Learning with Partitioning Reservoir Sampling

Chris Dongjoo Kim, Jinseo Jeong, Gunhee Kim

ECCV 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Classifier rebalancing and margin losses · 103 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 Partitioning Reservoir Sampling instead of conventional reservoir samplingreplay-based multi-label continual learning · COCOseq overallmAP: 50.2 -> 55.3Improved
use Partitioning Reservoir Sampling instead of conventional reservoir samplingreplay-based multi-label continual learning · NUS-WIDEseq overallmAP: 20.1 -> 25.5Improved
use PRS instead of CRS for minority classesreplay memory · COCOseq minority classesmAP: 26.9 -> 39.7Improved
use PRS instead of CRS for majority classesreplay memory · COCOseq majority classesmAP: 67.9 -> 67.5Hurt
use PRS instead of CRS with memory size 1000replay memory · COCOseq overallmAP: 47.1 -> 51.2Improved
use PRS instead of CRS with memory size 3000replay memory · COCOseq overallmAP: 51.1 -> 56.7Improved
use PRS instead of CRS under task schedule 2,3,0,1replay memory · COCOseq overallmAP: 48.2 -> 53.1Improved
use PRS instead of CRS under task schedule 3,1,0,2replay memory · COCOseq overallmAP: 49.7 -> 54.1Improved
move allocation power rho away from the balanced vicinity (ablation)PRS · COCOseqC-F1, O-F1, mAPHurt
add replay with PRS over plain finetuningonline multi-label learner · COCOseq overallmAP: 16.4 -> 55.3Improved
use PRS instead of ExStream, which uses task prior informationreplay memory · NUS-WIDEseq overallmAP: 21.0 -> 25.5Improved

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