AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Regulating Anatomy-Aware Rewards via Trajectory-Integral Feedback for Volumetric Computed Tomography Analysis

Tianwei Lin, Zhongwei Qiu, Jie Cao, Jiang Liu, Wenjie Yan, Zhang Bo, Yu Zhong, Wenqiao Zhang, Yingda Xia, Ling Zhang

ICML 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning · 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 TIF-GRPO RL stage after SFTQwen3-VL-4B · CT-RATE-Reportentity F1: 18.73 -> 39.11Improved
add TIF-GRPO RL stage after SFTQwen3-VL-4B · AMOS-MM-Reportentity F1: 17.65 -> 30.47Improved
scale TIF-GRPO model from 4B to 8BTIF-GRPO · CT-RATE-Reportentity F1: 39.11 -> 39.67No clear effect
replace ROUGE reward with trajectory-integral CABS reward (ablation)GRPO · CT-RATE-Reportentity F1: 29.74 -> 39.11Improved
replace RadGraph reward with TIF reward (ablation)GRPO · CT-RATE-Reportentity F1: 25.32 -> 39.11Improved
replace LLM rubric reward with TIF reward (ablation)GRPO · CT-RATE-Reportentity F1: 32.73 -> 39.11Improved
use unbalanced running cost vs control effort weights (0.4/1.0) (ablation)TIF-GRPO · CT-RATE-Reportentity F1: 39.11 -> 36.96Hurt
use unbalanced running cost vs control effort weights (1.0/0.4) (ablation)TIF-GRPO · CT-RATE-Reportentity F1: 39.11 -> 37.32Hurt
swap GRPO for RLOO, ReMAX or Reinforce++ with surface rewards (ablation)TIF-GRPO · CT-RATE-Reportentity F1: TIF-GRPO 39.11 vs best alternative 35.20Hurt
add TIF control on top of accuracy-reward GRPO (ablation)GRPO w/ Acc Reward (AMOS-MM) · CT-RATE-MCQ + AMOS-MM-MCQtotal average accuracy: 63.91 -> 69.67Improved
add TIF control on top of accuracy-reward GRPO (ablation)GRPO w/ Acc Reward (CT-RATE) · CT-RATE-MCQ + AMOS-MM-MCQtotal average accuracy: 63.59 -> 66.65Improved
apply TIF-GRPO to 2D chest X-ray report generationQwen3-VL · MIMIC-CXR-ReportF1: 25.3 -> 70.8Improved

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