AI Paper Atlas

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

Every Error has Its Magnitude: Asymmetric Mistake Severity Training for Multiclass Multiple Instance Learning

Sungrae Hong, Jiwon Jeong, Jisu Shin, Donghee Han, Sol Lee, Kyungeun Kim, Mun Yong Yi

CVPR 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 0 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 mistake severity cross-entropy L_MSCE (ablation)PAMS · BRACS (TransMIL)AsMC: 47.10 -> 55.65Improved
add mistake severity cross-entropy L_MSCE (ablation)PAMS · BRACS (DTFD-MIL-AFS)AsMC: 54.45 -> 54.64No clear effect
add hierarchy alignment loss L_HA (ablation)PAMS · BRACS (TransMIL)ACC: 45.21 -> 47.59Improved
add Semantic Feature Remix (SFR) (ablation)PAMS · BRACS (TransMIL)AsMC: 47.92 -> 55.65Improved
add Semantic Feature Remix (SFR) (ablation)PAMS · BRACS (TransMIL)ACC: 47.51 -> 47.59No clear effect
SFR instead of random remix (ablation)CE + TransMIL · In-house colonAsMC: 54.34 -> 55.86Improved
random or pseudo remix on in-house data (ablation)CE + TransMIL · In-house colonACC: 90.41 -> 89.25Hurt
add any remix strategy (ablation)CE + TransMIL · In-house colonAsMC: 44.16 -> 51.64Improved
PAMS vs prior priority learning (Hong et al.) on BRACSTransMIL · BRACSAsMC: 45.54 -> 55.65Improved
MSCE loss instead of CE on CIFAR-10 with class priorityResNet-18 · CIFAR-10 with class priorityAsMC: 34.84 -> 35.70Improved
class-distance weighted CE instead of CEResNet-18 · CIFAR-10 with class priorityAsMC: 34.84 -> 34.63No clear effect
weighted CE (1:10) instead of CEResNet-18 · CIFAR-10 with class priorityACC: 83.24 -> 73.69Hurt

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