AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Learning Multiple Pixelwise Tasks Based on Loss Scale Balancing

Jae-Han Lee, Chul Lee, Chang-Su Kim

ICCV 2021 · Learning theory and algorithms › Prediction and decision losses · approach: Ranking-based loss optimization · 23 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 LSB loss scale balancing instead of equal weightingmulti-column MTL, MobileNet.v2 · NYUv2average rank: 3.92 -> 1.00Improved
use LSB instead of equal weightingfull-sharing MTL, MobileNet.v2 · NYUv2average rank: 4.00 -> 1.50Improved
use LSB instead of equal weightingmulti-decoder MTL, MobileNet.v2 · NYUv2average rank: 2.92 -> 2.25Improved
use LSB instead of equal weightingfull-sharing MTL, PNASNet · NYUv2average rank: 3.17 -> 1.33Improved
use LSB instead of equal weightingmulti-column MTL, PNASNet · NYUv2average rank: 4.00 -> 1.33Improved
use LSB instead of equal weightingmulti-column MTL on Taskonomy 8-task · Taskonomyaverage rank: 3.56 -> 1.88Improved
use LSB instead of GLS for 2D edge lossmulti-column MTL on Taskonomy 8-task · Taskonomy2D edge loss: 0.119 -> 0.114Improved
use LSB instead of equal weighting for landmark localizationmulti-column MTL · CelebAeye pixel distance: 0.859 -> 0.641Improved
use LSB instead of equal weighting for attribute classificationmulti-column MTL · CelebAaccuracy: 90.6% -> 90.8%No clear effect
increase beta gradually from 0 (curriculum on task difficulty) instead of ignoring difficulty (beta = 0) (ablation)LSB · NYUv2 segmentationmIoU: 27.5% -> 32.6%Improved
use fixed beta = 1 (focus on difficult tasks) instead of beta = 0 (ablation)LSB · NYUv2 segmentationmIoU: 27.5% -> 28.1%Improved
use negative beta = -1 (focus on easy tasks) instead of beta = 0 (ablation)LSB · NYUv2 segmentationmIoU: 27.5% -> 25.5%Hurt

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