Xingjian Diao

About Me

I am currently a Ph.D. candidate at Dartmouth College. My research interests focus on multimodal learning. I have published papers and developed codes for temporal modeling, efficient training, and audio-video-language integration, advancing state-of-the-art multimodal large language models (MLLMs) applications such as multimodal question answering. During my Ph.D. studies, I interned at Amazon as an Applied Scientist in 2025.

Prior to Dartmouth, I earned a Master's degree in Computer Science from Northwestern University (2021), advised by Prof. Nabil Alshurafa (Thank you, Nabil!), and a B.S. degree in Computer Science from the University of Pittsburgh (2020).

Interests

Multimodal Large Language Models
Video Understanding
Natural Language and Speech Processing

Education

Ph.D. in Computer Science, -Present

Dartmouth College
M.S. in Computer Science, 2021

Northwestern University
B.S. in Computer Science, 2020

University of Pittsburgh

Publications

Selected Publications

	SoundMind: RL-Incentivized Logic Reasoning for Audio-Language Models Xingjian Diao, Chunhui Zhang, Keyi Kong, Weiyi Wu, Chiyu Ma, Zhongyu Ouyang, Peijun Qing, Soroush Vosoughi, Jiang Gui Conference on Empirical Methods in Natural Language Processing (EMNLP), 2025 (Oral Presentation)
	An End-to-End Adaptable Prototypical Framework for Explainable Fine-Grained Visual Question Answering Xingjian Diao^, Weiyi Wu^, Keyi Kong, Peijun Qing, Xinwen Xu, Ming Cheng, Soroush Vosoughi, Jiang Gui Conference on Empirical Methods in Natural Language Processing (EMNLP), 2025
	Knowing More, Acting Better: Hierarchical Representation for Embodied Decision-Making Chunhui Zhang, Zhongyu Ouyang, Xingjian Diao, Zheyuan Liu, Soroush Vosoughi Conference on Empirical Methods in Natural Language Processing (EMNLP 2025) Findings
	Assessing and Mitigating Medical Knowledge Drift and Conflicts in Large Language Models Weiyi Wu, Xinwen Xu, Chongyang Gao, Xingjian Diao, Siting Li, Lucas A Salas, Jiang Gui Conference on Empirical Methods in Natural Language Processing (EMNLP 2025) Findings
	Temporal Working Memory: Query-Guided Temporal Segment Refinement for Enhanced Multimodal Understanding Xingjian Diao^, Chunhui Zhang^, Weiyi Wu, Zhongyu Ouyang, Peijun Qing, Ming Cheng, Soroush Vosoughi, Jiang Gui Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL 2025) Findings 🎖 Guarini School of Graduate and Advanced Studies Travel Award, Dartmouth College
	Learning Sparsity for Effective and Efficient Music Performance Question Answering Xingjian Diao, Tianzhen Yang, Chunhui Zhang, Weiyi Wu, Ming Cheng, Jiang Gui Annual Meeting of the Association for Computational Linguistics (ACL), 2025
	FT2TF: First-Person Statement Text-To-Talking Face Generation Xingjian Diao, Ming Cheng, Wayner Barrios, SouYoung Jin Winter Conference on Applications of Computer Vision (WACV), 2025
	SPAN: Unlocking Pyramid Representations for Gigapixel Histopathological Images Weiyi Wu, Xingjian Diao, Chongyang Gao, Xinwen Xu, Siting Li, Jiang Gui arXiv, 2025
	On The Design Choices of Next Level LLMs Yijun Tian, Xingjian Diao, Ming Cheng, Chunhui Zhang, Jiang Gui, Soroush Vosoughi, Xiangliang Zhang, Nitesh V. Chawla, Shichao Pei arXiv, 2025
	Music's Multimodal Complexity in AVQA: Why We Need More than General Multimodal LLMs Wenhao You^, Xingjian Diao^, Chunhui Zhang, Keyi Kong, Weiyi Wu, Zhongyu Ouyang, Chiyu Ma, Tingxuan Wu, Noah Wei, Zong Ke, Ming Cheng, Soroush Vosoughi, Jiang Gui arXiv, 2025
	Judging the Judges: A Systematic Study of Position Bias in LLM-as-a-Judge Lin Shi, Chiyu Ma, Wenhua Liang, Xingjian Diao, Weicheng Ma, Soroush Vosoughi arXiv, 2025
	Learning Musical Representations for Music Performance Question Answering Xingjian Diao, Chunhui Zhang, Tingxuan Wu, Ming Cheng, Zhongyu Ouyang, Weiyi Wu, Jiang Gui Conference on Empirical Methods in Natural Language Processing (EMNLP 2024) Findings 🎖 Biomedical Data Science Travel Award, Dartmouth College
	AlphaExpert: Assigning LoRA Experts Based on Layer Training Quality Peijun Qing, Chongyang Gao, Yefan Zhou, Xingjian Diao, Yaoqing Yang, Soroush Vosoughi Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024
	GluMarker: A Novel Predictive Modeling of Glycemic Control Through Digital Biomarkers Ziyi Zhou^, Ming Cheng^, Xingjian Diao^, Yanjun Cui, Xiangling Li Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2024 (Lecture Presentation)* 🎖 IEEE EMBC NextGen Scholar Award
	An End-to-End Energy-Efficient Approach for Intake Detection With Low Inference Time Using Wrist-Worn Sensor Boyang Wei, Shibo Zhang, Xingjian Diao, Qiuyang Xu, Yang Gao, Nabil Alshurafa IEEE Journal of Biomedical and Health Informatics (JBHI), 2023

Projects

Selected Projects

	Intake Detection Tool with Multiple Classifiers An Android application for wrist-worn devices to detect feeding patterns with low energy consumption and fast inference times. It applied template-based multi-centroid classifier which could provide an end-to-end battery-efficient approach for feeding detection.
	Interactive Active Learning Annotation Tool An interactive annotation software that utilizes active learning to reduce data labeling time and cost. The front-end was created with PyQt5 and pyqtgraph, offering features such as time synchronization and video frame-by-frame rewinding. The back-end, utilizing cv2, sklearn and xgboost, performed data processing, K-means clustering, and clustered entropy active learning.
	iPADshiny iPADshiny (integrated Protein Array Data management,analysis and visualization tools) is a desktop application that simplifies protein analysis for biologists. It integrates multiple algorithms, including the auto-antibody Profiling Analysis, and utilizes state-of-the-art computational methods for efficient and effective analysis.
	Online Drawing Management System An Online Drawing Management System with B/S structure and Windows OS, including features such as notice announcement, navigation menu, user and role management, flexible authorization, and online management and preview of large drawing documents. It automatically loads existing document storage structures, eliminating the need for manual entry of basic information. (Copyright: 2018SR071476)
	Remote Voting System A remote voting system that uses SMS texts to count unique votes while recording phone numbers to prevent repetitive voting, offering an accessible and transparent solution for remote voting scenarios.
	Introvert An inclusive online chat environment for introverted students, utilizing JavaScript, Python, and Google Cloud platform to implement anonymous chatting and user-friendly direct messaging features, aimed at promoting engagement and improving the chat experience for introverted individuals.