AI Paper Atlas

Research map / Continual learning / Class-incremental

Fairness Continual Learning Approach to Semantic Scene Understanding in Open-World Environments

Thanh-Dat Truong, Hoang-Quan Nguyen, Bhiksha Raj, Khoa Luu

NeurIPS 2023 · Continual learning › Class-incremental · approach: Distillation-based continual segmentation · 5 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 prototypical contrastive clustering loss L_cluster (ablation)FairCL DeepLab-V3 · ADE20K 100-50mIoU all: 6.52 -> 32.97Improved
add prototypical contrastive clustering loss L_cluster (ablation)FairCL SegFormer · ADE20K 100-50mIoU all: 7.74 -> 36.18Improved
add fairness loss L_class (ablation)FairCL DeepLab-V3 · ADE20K 100-50mIoU all: 32.97 -> 34.40Improved
add fairness loss L_class (ablation)FairCL SegFormer · ADE20K 100-50mIoU all: 36.18 -> 36.78Improved
add fairness loss L_class (ablation)FairCL · ADE20K 100-50STD of IoU: reduced by 0.96% and 0.46%Improved
add conditional structural consistency loss L_cons (ablation)FairCL DeepLab-V3 · ADE20K 100-50mIoU all: 34.40 -> 36.99Improved
add conditional structural consistency loss L_cons (ablation)FairCL SegFormer · ADE20K 100-50 classes 101-150mIoU: 23.50 -> 25.46Improved
add L_cons (ablation)FairCL SegFormer · ADE20K 100-50 classes 0-100mIoU: 43.35 -> 43.56Improved
add L_cluster (ablation)FairCL · ADE20K 100-50 STDSTD of IoU: 19.90 -> 23.03Hurt
replace DeepLab-V3 ResNet101 with SegFormer MiT-B3FairCL · ADE20K 100-50mIoU classes 101-150: 19.86% to 25.46%Improved
vary margin Delta from 10 to 5 or 15 (ablation)FairCL · ADE20K 100-50mIoU all: 36.99 -> 36.68No clear effect
add L_class and L_cons to L_cluster (ablation)FairCL DeepLab-V3 · ADE20K minor groupmIoU: 25.13 -> 30.52Improved

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