Med-Banana: Learning Agentic Quality-Controlled Medical Image Editing from Success-and-Failure Trajectories

Zhihui Chen, Qingyuan Lei, Mengling Feng

Conference on Empirical Methods in Natural Language Processing (EMNLP) Accepted

Med-Banana connects my RSI research to medical agents through prompt-level self-improvement. An editor, verifier, and refiner form a recursive feedback loop: diagnose failed edits, revise positive and negative prompts using failure history, and retry from the original image. Success-and-failure trajectories supervise all three components, connecting agentic post-training with test-time refinement. Med-Banana-80K preserves 50,635 successful and 37,822 failed attempts across three imaging modalities and 23 disease categories.

Paper Code Dataset Project Page

Looking Ahead to Stay Grounded: Future-Discounted Visual Credit Assignment for VLM Reinforcement Learning

Zhihui Chen, Yike Yun, et al.

ICLR 2026 submission Under Review First Author

Farsight targets a core failure mode of RLVR post-training for vision-language models: token-level credit assignment that only rewards instant visual dependence, which lets models drift into ungrounded reasoning over long horizons. We extend credit assignment to future-discounted visual grounding — a token earns visual credit not only for the evidence it uses now, but for how that evidence supports discounted future reasoning steps — encouraging policies that stay anchored to the image across multi-step chains of thought. Developed during my foundation-model reasoning post-training research internship at ByteDance, on a 100×H100 cluster.

MedForge: Interpretable Medical Deepfake Detection via Forgery-aware Reasoning

Zhihui Chen, Kai He, Qingyuan Lei, Bin Pu, Jian Zhang, Yuling Xu, Mengling Feng^# (^# corresponding author 通讯作者)

Annual Meeting of the Association for Computational Linguistics (ACL) Main Conference

As generative models improve, medical deepfakes that implant or remove lesions while staying visually plausible pose growing risks to clinical safety and the integrity of medical evidence. Most prior work reduces detection to binary real-vs-fake scoring with little insight into where manipulation occurs or why. We present MedForge, an interpretable framework that introduces MedForge-90K—the first large-scale explainable medical deepfake dataset spanning CT, MRI, and X-ray, covering 19 lesion types with forgeries from 10 state-of-the-art deepfake models, each paired with expert-guided localization and clinical-grade explanations—and MedForge-Reasoner, a detector trained with a Localize-then-Analyze chain-of-thought paradigm and Forgery-aware GSPO reinforcement learning. MedForge-Reasoner achieves state-of-the-art detection while producing localized, verifiable medical rationales.

Paper Project Page Model Dataset

DivScore: Zero-Shot Detection of LLM-Generated Text in Specialized Domains

Zhihui Chen, Kai He, Yucheng Huang, Yunxiao Zhu, Mengling Feng

Conference on Empirical Methods in Natural Language Processing (EMNLP) Main Conference

Detecting LLM-generated text in specialized and high-stakes domains like medicine and law is crucial for combating misinformation and ensuring authenticity. We propose DivScore, a zero-shot detection framework using normalized entropy-based scoring and domain knowledge distillation to robustly identify LLM-generated text in specialized domains. Experiments show that DivScore consistently outperforms state-of-the-art detectors, with 14.4% higher AUROC and 64.0% higher recall at 0.1% false positive rate threshold.

Paper Code Project Page

Production Efficiency Analysis Based on RFID-Collected Manufacturing Big Data

Zhihui Chen

52nd NAMRC (North American Manufacturing Research Conference)

This paper presents a production efficiency analysis framework based on RFID-collected manufacturing big data, enabling real-time monitoring and optimization of manufacturing processes.

Clustering Enabled Few-Shot Load Forecasting

Qiyuan Wang, Zhihui Chen

IEEE iSPEC (International Smart Power and Energy Conference)

We propose a clustering-enabled few-shot learning approach for load forecasting, which significantly improves prediction accuracy with limited training data. This work also led to a patent (CN113887812B).