AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Towards Reliable Medical Large Vision-Language Models via Counterfactual Preference Optimization

Xiaoguang Zhu, NaipengWang NaipengWang, Kartik Patwari, Lianlong Sun, Chen-Nee Chuah, ChengxinPang ChengxinPang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Bias-aware multi-image robustness · 1 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 standard DPO loss L_DPO alongside counterfactual loss (ablation)CoMedPO (SFT) · VQA-RAD Openrecall: 29.50 -> 42.50Improved
use counterfactual loss instead of causal-weighted DPO loss (ablation)CoMedPO (SFT) · VQA-RAD Openrecall: 37.50 -> 42.50Improved
add counterfactual clinical query for composite pairs (ablation)CoMedPO (SFT) · VQA-RAD Openrecall: 42.00 -> 42.50No clear effect
use matched prompt with composite image at inference instead of composite only (ablation)CoMedPO (SFT) · VQA-RAD Openrecall: 37.50 -> 43.43Improved
use causally constructed background instead of random patch in composite (ablation)CoMedPO (SFT) · VQA-RAD Openrecall: 37.00 -> 42.50Improved
add CoMedPO counterfactual optimization to MMedPO (no SFT)MMedPO · SLAKE Closedaccuracy: 58.25 -> 71.17Improved
add CoMedPO to MMedPO (no SFT)MMedPO · IU-XRayROUGE-L: 29.52 -> 32.68Improved
add CoMedPO to SFT-enhanced MMedPOMMedPO + SFT · SLAKE Closedaccuracy: 67.82 -> 74.40Improved
add CoMedPO to SFT-enhanced MMedPOMMedPO + SFT · VQA-RAD Closedaccuracy: 67.73 -> 74.90Improved
add counterfactual optimization to mDPOmDPO · SLAKE Closedaccuracy: 60.29 -> 65.07Improved
add counterfactual optimization to mDPOmDPO · IU-XRayROUGE-L: 22.38 -> 30.65Improved
add counterfactual optimization to SPPOSPPO · VQA-RAD Openrecall: 38.50 -> 42.50Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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