Publications

2025

1. AceSearcher: Bootstrapping Reasoning and Search for LLMs via Reinforced Self-Play

   Ran Xu, Yuchen Zhuang, Zihan Dong, Jonathan Wang, Yue Yu, Joyce C. Ho, Linjun Zhang, Haoyu Wang, Wenqi Shi, and Carl Yang

   Proceedings of NeurIPS, 2025. (Spotlight)

   arXiv Code Huggingface
2. Retrieval-augmented GUI Agents with Generative Guidelines

   Ran Xu, Kaixin Ma, Wenhao Yu, Hongming Zhang, Joyce C. Ho, Carl Yang, and Dong Yu

   Proceedings of EMNLP, 2025.

   arXiv
3. RAG in the Wild: On the (In) effectiveness of LLMs with Mixture-of-Knowledge Retrieval Augmentation

   Ran Xu, Yuchen Zhuang, Yue Yu, Haoyu Wang, Wenqi Shi, and Carl Yang

   arXiv preprint arXiv:2507.20059, 2025.

   arXiv Code Huggingface
4. MedAgentGym: Training LLM Agents for Code-Based Medical Reasoning at Scale

   Ran Xu*, Yuchen Zhuang*, Yishan Zhong, Yue Yu, Xiangru Tang, Hang Wu, May D Wang, Peifeng Ruan, Donghan Yang, Tao Wang, Guanghua Xiao, Carl Yang, Yang Xie, and Wenqi Shi

   arXiv preprint arXiv:2506.04405, 2025.

   arXiv Code
5. RoseRAG: Robust Retrieval-augmented Generation with Small-scale LLMs via Margin-aware Preference Optimization

   Tianci Liu, Haoxiang Jiang, Tianze Wang, Ran Xu, Yue Yu, Linjun Zhang, Tuo Zhao, and Haoyu Wang

   Proceedings of ACL (Findings), 2025.

   Abs arXiv

   Large language models (LLMs) have achieved impressive performance but face high computational costs and latency, limiting their deployment in resource-constrained settings. In contrast, small-scale LLMs (SLMs) are more efficient yet struggle to capture evolving real-world knowledge. Retrieval-augmented generation (RAG) helps by integrating external knowledge, but imperfect retrieval can introduce distracting noise that misleads SLMs. We propose RoseRAG, a robust RAG framework for SLMs via Margin-aware Preference Optimization. RoseRAG employs multi-turn prompting for detailed reasoning, rejection sampling for high-quality explanations, and contrastive preference selection to refine responses by maximizing the likelihood gap between preferred and non-preferred outputs. By integrating these components into a margin-aware optimization process, RoseRAG robustly enhances the accuracy and reliability of SLMs for RAG applications. Extensive experiments on three open-domain question answering benchmarks indicate that our innovative RoseRAG surpasses state-of-the-art baselines significantly.

6. Collab-RAG: Boosting Retrieval-Augmented Generation for Complex Question Answering via White-Box and Black-Box LLM Collaboration

   Ran Xu, Wenqi Shi, Yuchen Zhuang, Yue Yu, Joyce C. Ho, Haoyu Wang, and Carl Yang

   Proceedings of COLM, 2025.

   arXiv Code
7. SimRAG: Self-Improving Retrieval-Augmented Generation for Adapting Large Language Models to Specialized Domains

   Ran Xu, Hui Liu, Sreyashi Nag, Zhenwei Dai, Yaochen Xie, Xianfeng Tang, Chen Luo, Yang Li, Joyce C. Ho, Carl Yang, and Qi He

   Proceedings of NAACL, 2025.

   Abs arXiv

   Retrieval-augmented generation (RAG) enhances the question answering (QA) abilities of large language models (LLMs) by integrating external knowledge. However, adapting general-purpose RAG systems to specialized fields such as science and medicine poses unique challenges due to distribution shifts and limited access to domain-specific data. To tackle this, we propose SimRAG, a self-training approach that equips LLMs with joint capabilities of question answering and question generation for domain adaptation. Our method first fine-tunes LLMs on instruction-following, question-answering, and search-related data. Then, it prompts LLMs to generate diverse domain-relevant questions from unlabeled corpora, with an additional filtering strategy to retain high-quality synthetic examples. By leveraging these synthetic examples, the LLMs can improve their performance on domain-specific RAG tasks. Experiments on 11 datasets across three different domains verify the efficacy of SimRAG over baselines by 1.2%–8.6%.

2024

1. Boosting reward model with preference-conditional multi-aspect synthetic data generation

   Jiaming Shen, Ran Xu, Yennie Jun, Zhen Qin, Tianqi Liu, Carl Yang, Yi Liang, Simon Baumgartner, and Michael Bendersky

   arXiv preprint arXiv:2407.16008, 2024.

   arXiv
2. BMRetriever: Tuning Large Language Models as Better Biomedical Text Retrievers

   Ran Xu*, Wenqi Shi*, Yue Yu*, Yuchen Zhuang, Yanqiao Zhu, May Dongmei Wang, Joyce C. Ho, Chao Zhang, and Carl Yang

   Proceedings of EMNLP, 2024.

   Abs arXiv Code Huggingface

   Developing effective biomedical retrieval models is important for excelling at knowledge-intensive biomedical tasks but still challenging due to the lack of sufficient publicly annotated biomedical data and computational resources. We present BMRetriever, a series of dense retrievers for enhancing biomedical retrieval via unsupervised pre-training on large biomedical corpora, followed by instruction fine-tuning on a combination of labeled datasets and synthetic pairs. Experiments on 5 biomedical tasks across 11 datasets verify BMRetriever’s efficacy on various biomedical applications. BMRetriever also exhibits strong parameter efficiency, with the 410M variant outperforming baselines up to 11.7 times larger, and the 2B variant matching the performance of models with over 5B parameters. The training data and model checkpoints are released at https://huggingface.co/BMRetriever to ensure transparency, reproducibility, and application to new domains.

3. MedAdapter: Efficient Test-Time Adaptation of Large Language Models Towards Medical Reasoning

   Wenqi Shi*, Ran Xu*, Yuchen Zhuang, Yue Yu, Haotian Sun, Hang Wu, Carl Yang, and May Dongmei Wang

   Proceedings of EMNLP, 2024.

   Abs arXiv Code

   Despite their improved capabilities in generation and reasoning, adapting large language models (LLMs) to the biomedical domain remains challenging due to their immense size and privacy concerns. In this study, we propose MedAdapter, a unified post-hoc adapter for test-time adaptation of LLMs towards biomedical applications. Instead of fine-tuning the entire LLM, MedAdapter effectively adapts the original model by fine-tuning only a small BERT-sized adapter to rank candidate solutions generated by LLMs. Experiments on four biomedical tasks across eight datasets demonstrate that MedAdapter effectively adapts both white-box and black-box LLMs in biomedical reasoning, achieving average performance improvements of 18.24% and 10.96%, respectively, without requiring extensive computational resources or sharing data with third parties. MedAdapter also yields enhanced performance when combined with train-time adaptation, highlighting a flexible and complementary solution to existing adaptation methods. Faced with the challenges of balancing model performance, computational resources, and data privacy, MedAdapter provides an efficient, privacy-preserving, cost-effective, and transparent solution for adapting LLMs to the biomedical domain.

4. EHRAgent: Code Empowers Large Language Models for Few-shot Complex Tabular Reasoning on Electronic Health Records

   Wenqi Shi*, Ran Xu*, Yuchen Zhuang, Yue Yu, Jieyu Zhang, Hang Wu, Yuanda Zhu, Joyce C. Ho, Carl Yang, and May Dongmei Wang

   Proceedings of EMNLP, 2024.

   Abs arXiv Code

   Clinicians often rely on data engineers to retrieve complex patient information from electronic health record (EHR) systems, a process that is both inefficient and time-consuming. We propose EHRAgent, a large language model (LLM) agent empowered with accumulative domain knowledge and robust coding capability. EHRAgent enables autonomous code generation and execution to facilitate clinicians in directly interacting with EHRs using natural language. Specifically, we formulate a multi-tabular reasoning task based on EHRs as a tool-use planning process, efficiently decomposing a complex task into a sequence of manageable actions with external toolsets. We first inject relevant medical information to enable EHRAgent to effectively reason about the given query, identifying and extracting the required records from the appropriate tables. By integrating interactive coding and execution feedback, EHRAgent then effectively learns from error messages and iteratively improves its originally generated code. Experiments on three real-world EHR datasets show that EHRAgent outperforms the strongest baseline by up to 29.6% in success rate, verifying its strong capacity to tackle complex clinical tasks with minimal demonstrations.

5. Knowledge-Infused Prompting: Assessing and Advancing Clinical Text Data Generation with Large Language Models

   Ran Xu, Hejie Cui, Yue Yu, Xuan Kan, Wenqi Shi, Yuchen Zhuang, May Dongmei Wang, Wei Jin, Joyce Ho, and Carl Yang

   Proceedings of ACL (Findings), 2024.

   Abs arXiv Code

   Clinical natural language processing faces challenges like complex medical terminology and clinical contexts. Recently, large language models (LLMs) have shown promise in this domain. Yet, their direct deployment can lead to privacy issues and are constrained by resources. To address this challenge, we delve into synthetic clinical text generation with LLMs for clinical NLP tasks. We propose an innovative, resource-efficient approach, ClinGen, which infuses knowledge into the process. Our model involves clinical knowledge extraction and context-informed LLM prompting. Both clinical topics and writing styles are drawn from external domain-specific knowledge graphs and LLMs to guide data generation. Our extensive empirical study across 8 clinical NLP tasks and 18 datasets reveals that ClinGen consistently enhances performance across various tasks by 7.7%-8.7% on average, effectively aligning the distribution of real datasets and enriching the diversity of generated training instances.

6. RAM-EHR: Retrieval Augmentation Meets Clinical Predictions on Electronic Health Records

   Ran Xu, Wenqi Shi, Yue Yu, Yuchen Zhuang, Bowen Jin, May Dongmei Wang, Joyce Ho, and Carl Yang

   Proceedings of ACL, 2024. (Oral)

   Abs arXiv Code

   We present RAM-EHR, a Retrieval AugMentation pipeline to improve clinical predictions on Electronic Health Records (EHRs). RAM-EHR first collects multiple knowledge sources, converts them into text format, and uses dense retrieval to obtain information related to medical concepts. This strategy addresses the difficulties associated with complex names for the concepts. RAM-EHR then augments the local EHR predictive model co-trained with consistency regularization to capture complementary information from patient visits and summarized knowledge. Experiments on two EHR datasets show the efficacy of RAM-EHR over previous knowledge-enhanced baselines (3.4% gain in AUROC and 7.2% gain in AUPR), emphasizing the effectiveness of the summarized knowledge from RAM-EHR for clinical prediction tasks.

7. From Basic to Extra Features: Hypergraph Transformer Pretrain-then-Finetuning for Balanced Clinical Predictions on EHR

   Ran Xu, Yiwen Lu, Chang Liu, Yong Chen, Yan Sun, Xiao Hu, Joyce C Ho, and Carl Yang

   Proceedings of CHIL, 2024.

   Abs arXiv PDF

   Electronic Health Records (EHRs) contain rich patient information and are crucial for clinical research and practice. In recent years, deep learning models have been applied to EHRs, but they often rely on massive features, which may not be readily available for all patients. We propose \ours\footnoteShort for \textbfHypergraph \textbfTransformer \textbfPretrain-then-Finetuning with \textbfSmoo\textbfthness-induced regularization \textbfand \textbfReweighting., which leverages hypergraph structures with a pretrain-then-finetune framework for modeling EHR data, enabling seamless integration of additional features. Additionally, we design two techniques, namely (1) \emphSmoothness-inducing Regularization and (2) \emphGroup-balanced Reweighting, to enhance the model’s robustness during finetuning. Through experiments conducted on two real EHR datasets, we demonstrate that \ours consistently outperforms various baselines while striking a balance between patients with basic and extra features.

8. TACCO: Task-guided Co-clustering of Clinical Concepts and Patient Visits for Disease Subtyping based on EHR Data

   Ziyang Zhang, Hejie Cui, Ran Xu, Yuzhang Xie, Joyce C Ho, and Carl Yang

   Proceedings of KDD, 2024.

   arXiv PDF

2023

1. Weakly-supervised Scientific Document Classification via Retrieval-Augmented Multi-stage Training

   Ran Xu*, Yue Yu*, Joyce C Ho, and Carl Yang

   Proceedings of SIGIR, 2023.

   arXiv Code
2. Neighborhood-regularized Self-Training for Learning with Few Labels

   Ran Xu, Yue Yu, Hejie Cui, Xuan Kan, Yanqiao Zhu, Joyce Ho, Chao Zhang, and Carl Yang

   Proceedings of AAAI, 2023. (Oral)

   arXiv PDF Code
3. Hypergraph Transformers for EHR-based Clinical Predictions

   Ran Xu, Mohammed K Ali, Joyce C Ho, and Carl Yang

   AMIA Summits on Translational Science Proceedings, 2023. (Oral)

   PDF
4. Open Visual Knowledge Extraction via Relation-Oriented Multimodality Model Prompting

   Hejie Cui, Xinyu Fang, Zihan Zhang, Ran Xu, Xuan Kan, Xin Liu, Yue Yu, Manling Li, Yangqiu Song, and Carl Yang

   Proceedings of NeurIPS, 2023.

   arXiv PDF
5. Cold-Start Data Selection for Better Few-shot Language Model Fine-tuning: A Prompt-based Uncertainty Propagation Approach

   Yue Yu, Rongzhi Zhang, Ran Xu, Jieyu Zhang, Jiaming Shen, and Chao Zhang

   Proceedings of ACL, 2023.

   Abs PDF Code

   We present PATRON, a prompt-based data selection method for pre-trained language model fine-tuning under cold-start scenarios, i.e., no initial labeled data are available. In PATRON, we design (1) a prompt-based uncertainty propagation approach to estimate the importance of data points and (2) a partition-then-rewrite (PTR) strategy to promote sample diversity when querying for annotations. Experiments on six text classification datasets show that PATRON outperforms the strongest cold-start data selection baselines by up to 6.9%. Besides, with 128 labels only, PATRON achieves 91.0% and 92.1% of the fully supervised performance based on vanilla fine-tuning and prompt-based learning respectively. Our implementation of PATRON will be published upon acceptance.

6. R-Mixup: Riemannian Mixup for Biological Networks

   Xuan Kan, Zimu Li, Hejie Cui, Yue Yu, Ran Xu, Shaojun Yu, Zilong Zhang, Ying Guo, and Carl Yang

   Proceedings of KDD, 2023.

   arXiv PDF
7. Deep DAG Learning on Brain Networks for fMRI Analysis

   Yue Yu, Xuan Kan, Hejie Cui, Ran Xu, Yujia Zheng, Xiangchen Song, Yanqiao Zhu, Kun Zhang, Razieh Nabi, Ying Guo, Chao Zhang, and Carl Yang

   Proceedings of ISBI, 2023.

   arXiv PDF
8. A Survey on Knowledge Graphs for Healthcare: Resources, Application Progress, and Promise

   Hejie Cui, Jiaying Lu, Shiyu Wang, Ran Xu, Wenjing Ma, Shaojun Yu, Yue Yu, Xuan Kan, Tianfan Fu, Chen Ling, Joyce Ho, Fei Wang, and Carl Yang

   Proceedings of ICML (IMLH Workshop), 2023.

   arXiv PDF

2022

1. Counterfactual and Factual Reasoning over Hypergraphs for Interpretable Clinical Predictions on EHR

   Ran Xu, Yue Yu, Chao Zhang, Mohammed K Ali, Joyce C Ho, and Carl Yang

   Proceedings of ML4H, 2022. (Best Paper Award)

   Abs PDF Code

   Electronic Health Record modeling is crucial for digital medicine. However, existing models ignore higher-order interactions among medical codes and their causal relations towards downstream clinical predictions. To address such limitations, we propose a novel framework CACHE, to provide effective and insightful clinical predictions based on hypergraph representation learning and counterfactual and factual reasoning techniques. Experiments on two real EHR datasets show the superior performance of CACHE. Case studies with a domain expert illustrate a preferred capability of CACHE in generating clinically meaningful interpretations towards the correct predictions.