Rui Hou
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Rui Hou has a total of 125 co-authors. Here are the top co-authors by number of shared publications:
Jeffrey R. Marks
Department of Radiology, Duke University School of Medicine, 2424 Erwin Rd, Hock Plaza #302, Durham, NC 27710
6 shared publicationsAmjad Almahairi
University of California Los Angeles ‡ Meta AI
5 shared publicationsE. Shelley Hwang
Department of Radiology, Duke University School of Medicine, 2424 Erwin Rd, Hock Plaza #302, Durham, NC 27710
4 shared publications- Joseph Y. Lo
Duke University
4 shared publications - Lars J. Grimm
Department of Radiology, Duke University , Durham, NC
4 shared publications - Alexander Min
KAIST
3 shared publications
Latest Publications
- Open Access
- 18 August 2026
- Published by Frontiers Media SA in Frontiers in Plant Science
- Vol. 17, 1912355
Knowledge-enhanced large language model construction for garlic cultivation using GraphRAG
Introduction: Rapid advances in smart agriculture have highlighted the potential of large language models (LLMs), while practical applications remain limited by privacy risks, high training costs, and hallucinations. Focusing on garlic cultivation, this study proposes a knowledge-graph-enhanced framework that improves domain-specific LLMs through graph-based retrieval. Methods: A hybrid clustering algorithm with intra-cluster multidimensional ranking is employed to identify ten core entity types from garlic-related corpora. Few-shot learning and chain-of-thought prompting are further integrated to optimize entity and relation extraction using Qwen 2.5:32B, leading to the construction of a structured garlic cultivation knowledge graph. The knowledge graph is integrated with LLMs via GraphRAG, while prompt reconstruction strategies are adopted to enrich contextual information and constrain generation, thereby improving domain grounding and reducing hallucinations. Results: Experimental results demonstrate absolute improvements of 13.0, 37.0, and 28.2 percentage points in precision, recall, and F1-score, respectively, for triple extraction compared with the baseline LLM, along with a 37.5 percentage point increase in retrieval-based question-answering accuracy. Discussion: The proposed approach is interpretable and locally deployable without dependence on commercial APIs, offering a transferable methodological reference for domain-specific knowledge graph construction and intelligent question answering across diverse crop systems.
- 18 August 2026
- Published by Frontiers Media SA in Frontiers in Plant Science
- Vol. 17, 1912355
Open AccessKnowledge-enhanced large language model construction for garlic cultivation using GraphRAG
Introduction: Rapid advances in smart agriculture have highlighted the potential of large language models (LLMs), while practical applications remain limited by privacy risks, high training costs, and hallucinations. Focusing on garlic cultivation, this study proposes a knowledge-graph-enhanced framework that improves domain-specific LLMs through graph-based retrieval. Methods: A hybrid clustering algorithm with intra-cluster multidimensional ranking is employed to identify ten core entity types from garlic-related corpora. Few-shot learning and chain-of-thought prompting are further integrated to optimize entity and relation extraction using Qwen 2.5:32B, leading to the construction of a structured garlic cultivation knowledge graph. The knowledge graph is integrated with LLMs via GraphRAG, while prompt reconstruction strategies are adopted to enrich contextual information and constrain generation, thereby improving domain grounding and reducing hallucinations. Results: Experimental results demonstrate absolute improvements of 13.0, 37.0, and 28.2 percentage points in precision, recall, and F1-score, respectively, for triple extraction compared with the baseline LLM, along with a 37.5 percentage point increase in retrieval-based question-answering accuracy. Discussion: The proposed approach is interpretable and locally deployable without dependence on commercial APIs, offering a transferable methodological reference for domain-specific knowledge graph construction and intelligent question answering across diverse crop systems.
- 1 November 2025
- Published by Radiological Society of North America (RSNA) in Radiology
- Vol. 317 (2) , e243739
Cross-National Radiomics Validation Using Mammography to Predict Occult Invasion in Ductal Carcinoma in Situ
Mammographic radiomic features were able to predict occult invasive disease in core-needle biopsy–proven ductal carcinoma in situ in patients from three countries.
- 1 November 2025
- Published by Radiological Society of North America (RSNA) in Radiology
- Vol. 317 (2) , e243739
Cross-National Radiomics Validation Using Mammography to Predict Occult Invasion in Ductal Carcinoma in Situ
Mammographic radiomic features were able to predict occult invasive disease in core-needle biopsy–proven ductal carcinoma in situ in patients from three countries.
- Top Cited PapersOpen Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 11499-11514
Self-Generated Critiques Boost Reward Modeling for Language Models
Yue Yu, Zhengxing Chen, Aston Zhang, Liang Tan, Chenguang Zhu, Richard Yuanzhe Pang, Yundi Qian, Xuewei Wang, Suchin Gururangan, Chao Zhang, Melanie Kambadur, Dhruv Mahajan, Rui Hou. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 11499-11514
Top Cited PapersOpen AccessSelf-Generated Critiques Boost Reward Modeling for Language Models
Yue Yu, Zhengxing Chen, Aston Zhang, Liang Tan, Chenguang Zhu, Richard Yuanzhe Pang, Yundi Qian, Xuewei Wang, Suchin Gururangan, Chao Zhang, Melanie Kambadur, Dhruv Mahajan, Rui Hou. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- 12 April 2024
- Published by Elsevier in Laboratory Investigation
- Vol. 104 (6) , 102059
N6-Methyladenosine–Modified KREMEN2 Promotes Tumorigenesis and Malignant Progression of High-Grade Serous Ovarian Cancer
- 12 April 2024
- Published by Elsevier in Laboratory Investigation
- Vol. 104 (6) , 102059
N6-Methyladenosine–Modified KREMEN2 Promotes Tumorigenesis and Malignant Progression of High-Grade Serous Ovarian Cancer
- Open Access
- 7 February 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (2) , e0282402
Classification performance bias between training and test sets in a limited mammography dataset
Objectives: To assess the performance bias caused by sampling data into training and test sets in a mammography radiomics study.Methods: Mammograms from 700 women were used to study upstaging of ductal carcinoma in situ. The dataset was repeatedly shuffled and split into training (n = 400) and test cases (n = 300) forty times. For each split, cross-validation was used for training, followed by an assessment of the test set. Logistic regression with regularization and support vector machine were used as the machine learning classifiers. For each split and classifier type, multiple models were created based on radiomics and/or clinical features.Results: Area under the curve (AUC) performances varied considerably across the different data splits (e.g., radiomics regression model: train 0.58–0.70, test 0.59–0.73). Performances for regression models showed a tradeoff where better training led to worse testing and vice versa. Cross-validation over all cases reduced this variability, but required samples of 500+ cases to yield representative estimates of performance.Conclusions: In medical imaging, clinical datasets are often limited to relatively small size. Models built from different training sets may not be representative of the whole dataset. Depending on the selected data split and model, performance bias could lead to inappropriate conclusions that might influence the clinical significance of the findings.Advances in knowledge: Performance bias can result from model testing when using limited datasets. Optimal strategies for test set selection should be developed to ensure study conclusions are appropriate.
- 7 February 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (2) , e0282402
Open AccessClassification performance bias between training and test sets in a limited mammography dataset
Objectives: To assess the performance bias caused by sampling data into training and test sets in a mammography radiomics study.Methods: Mammograms from 700 women were used to study upstaging of ductal carcinoma in situ. The dataset was repeatedly shuffled and split into training (n = 400) and test cases (n = 300) forty times. For each split, cross-validation was used for training, followed by an assessment of the test set. Logistic regression with regularization and support vector machine were used as the machine learning classifiers. For each split and classifier type, multiple models were created based on radiomics and/or clinical features.Results: Area under the curve (AUC) performances varied considerably across the different data splits (e.g., radiomics regression model: train 0.58–0.70, test 0.59–0.73). Performances for regression models showed a tradeoff where better training led to worse testing and vice versa. Cross-validation over all cases reduced this variability, but required samples of 500+ cases to yield representative estimates of performance.Conclusions: In medical imaging, clinical datasets are often limited to relatively small size. Models built from different training sets may not be representative of the whole dataset. Depending on the selected data split and model, performance bias could lead to inappropriate conclusions that might influence the clinical significance of the findings.Advances in knowledge: Performance bias can result from model testing when using limited datasets. Optimal strategies for test set selection should be developed to ensure study conclusions are appropriate.
- Open Access
Co-training and Co-distillation for Quality Improvement and Compression of Language Models
Knowledge Distillation (KD) compresses computationally expensive pre-trained language models (PLMs) by transferring their knowledge to smaller models, allowing their use in resource-constrained or real-time settings. However, most smaller models fail to surpass the performance of the original larger model, resulting in sacrificing performance to improve inference speed. To address this issue, we propose Co-Training and Co-Distillation (CTCD), a novel framework that improves performance and inference speed together by co-training two models while mutually distilling knowledge. The CTCD framework successfully achieves this based on two significant findings: 1) Distilling knowledge from the smaller model to the larger model during co-training improves the performance of the larger model. 2) The enhanced performance of the larger model further boosts the performance of the smaller model. The CTCD framework shows promise as it can be combined with existing techniques like architecture design or data augmentation, replacing one-way KD methods, to achieve further performance improvement. Extensive ablation studies demonstrate the effectiveness of CTCD, and the small model distilled by CTCD outperforms the original larger model by a significant margin of 1.66 on the GLUE benchmark.
Open AccessCo-training and Co-distillation for Quality Improvement and Compression of Language Models
Knowledge Distillation (KD) compresses computationally expensive pre-trained language models (PLMs) by transferring their knowledge to smaller models, allowing their use in resource-constrained or real-time settings. However, most smaller models fail to surpass the performance of the original larger model, resulting in sacrificing performance to improve inference speed. To address this issue, we propose Co-Training and Co-Distillation (CTCD), a novel framework that improves performance and inference speed together by co-training two models while mutually distilling knowledge. The CTCD framework successfully achieves this based on two significant findings: 1) Distilling knowledge from the smaller model to the larger model during co-training improves the performance of the larger model. 2) The enhanced performance of the larger model further boosts the performance of the smaller model. The CTCD framework shows promise as it can be combined with existing techniques like architecture design or data augmentation, replacing one-way KD methods, to achieve further performance improvement. Extensive ablation studies demonstrate the effectiveness of CTCD, and the small model distilled by CTCD outperforms the original larger model by a significant margin of 1.66 on the GLUE benchmark.
- Open Access
Learning Easily Updated General Purpose Text Representations with Adaptable Task-Specific Prefixes
Many real-world applications require making multiple predictions from the same text. Fine-tuning a large pre-trained language model for each downstream task causes computational burdens in the inference time due to several times of forward passes. To amortize the computational cost, freezing the language model and building lightweight models for downstream tasks based on fixed text representations are common solutions. Accordingly, how to learn fixed but general text representations that can generalize well to unseen downstream tasks becomes a challenge. Previous works have shown that the generalizability of representations can be improved by fine-tuning the pre-trained language model with some source tasks in a multi-tasking way. In this work, we propose a prefix-based method to learn the fixed text representations with source tasks. We learn a task-specific prefix for each source task independently and combine them to get the final representations. Our experimental results show that prefix-based training performs better than multi-tasking training and can update the text representations at a smaller computational cost than multi-tasking training.
Open AccessLearning Easily Updated General Purpose Text Representations with Adaptable Task-Specific Prefixes
Many real-world applications require making multiple predictions from the same text. Fine-tuning a large pre-trained language model for each downstream task causes computational burdens in the inference time due to several times of forward passes. To amortize the computational cost, freezing the language model and building lightweight models for downstream tasks based on fixed text representations are common solutions. Accordingly, how to learn fixed but general text representations that can generalize well to unseen downstream tasks becomes a challenge. Previous works have shown that the generalizability of representations can be improved by fine-tuning the pre-trained language model with some source tasks in a multi-tasking way. In this work, we propose a prefix-based method to learn the fixed text representations with source tasks. We learn a task-specific prefix for each source task independently and combine them to get the final representations. Our experimental results show that prefix-based training performs better than multi-tasking training and can update the text representations at a smaller computational cost than multi-tasking training.
- Open Access
Query Rewriting for Effective Misinformation Discovery
We propose a novel system to help fact-checkers formulate search queries for known misinformation claims and effectively search across multiple social media platforms. We introduce an adaptable rewriting strategy, where editing actions for queries containing claims (e.g., swap a word with its synonym; change verb tense into present simple) are automatically learned through offline reinforcement learning. Our model uses a decision transformer to learn a sequence of editing actions that maximizes query retrieval metrics such as mean average precision. We conduct a series of experiments showing that our query rewriting system achieves a relative increase in the effectiveness of the queries of up to 42%, while producing editing action sequences that are human interpretable.
Open AccessQuery Rewriting for Effective Misinformation Discovery
We propose a novel system to help fact-checkers formulate search queries for known misinformation claims and effectively search across multiple social media platforms. We introduce an adaptable rewriting strategy, where editing actions for queries containing claims (e.g., swap a word with its synonym; change verb tense into present simple) are automatically learned through offline reinforcement learning. Our model uses a decision transformer to learn a sequence of editing actions that maximizes query retrieval metrics such as mean average precision. We conduct a series of experiments showing that our query rewriting system achieves a relative increase in the effectiveness of the queries of up to 42%, while producing editing action sequences that are human interpretable.
- Open Access
A Study on Knowledge Distillation from Weak Teacher for Scaling Up Pre-trained Language Models
Distillation from Weak Teacher (DWT) is a method of transferring knowledge from a smaller, weaker teacher model to a larger student model to improve its performance. Previous studies have shown that DWT can be effective in the vision domain and natural language processing (NLP) pre-training stage. Specifically, DWT shows promise in practical scenarios, such as enhancing new generation or larger models using pre-trained yet older or smaller models and lacking a resource budget. However, the optimal conditions for using DWT have yet to be fully investigated in NLP pre-training. Therefore, this study examines three key factors to optimize DWT, distinct from those used in the vision domain or traditional knowledge distillation. These factors are: (i) the impact of teacher model quality on DWT effectiveness, (ii) guidelines for adjusting the weighting value for DWT loss, and (iii) the impact of parameter remapping as a student model initialization technique for DWT.
Open AccessA Study on Knowledge Distillation from Weak Teacher for Scaling Up Pre-trained Language Models
Distillation from Weak Teacher (DWT) is a method of transferring knowledge from a smaller, weaker teacher model to a larger student model to improve its performance. Previous studies have shown that DWT can be effective in the vision domain and natural language processing (NLP) pre-training stage. Specifically, DWT shows promise in practical scenarios, such as enhancing new generation or larger models using pre-trained yet older or smaller models and lacking a resource budget. However, the optimal conditions for using DWT have yet to be fully investigated in NLP pre-training. Therefore, this study examines three key factors to optimize DWT, distinct from those used in the vision domain or traditional knowledge distillation. These factors are: (i) the impact of teacher model quality on DWT effectiveness, (ii) guidelines for adjusting the weighting value for DWT loss, and (iii) the impact of parameter remapping as a student model initialization technique for DWT.