About Me [CV]
I am a Ph.D. student in the SMILE Lab of the Department of ECE, Northeastern University, under the supervision of Prof. Yun Raymond Fu (Member of Academia Europaea, Fellow of ACM, AAAS, AAAI, IEEE, IAPR). I received my B.S. and M.S. degrees from Xidian University, advised by Prof. Xuefeng Liang. I have interned at Adobe Research.
Research interests Multimodal LLMs | Efficiency | Reliability | Hallucination Detection & Mitigation | Video Understanding | Layout Understanding
Actively seeking internship opportunities for Spring/Summer 2027. Feel free to reach out for collaborations or any inquiries!
News
TOP Paper list of Video LLM hallucination. Welcome to Star and Contribute!
Jun. 2026 One paper accepted by ECCV 2026 (Adobe internship project). Thanks to my mentors! May 2026 Started my internship at Adobe Research, excited to work with Zhaowen again. Apr. 2026 One Video-LLM Hallucination survey paper accepted by ACL 2026 Apr. 2026 One co-authored paper IDEA accepted by IEEE FG 2026 Jan. 2026 One paper SHIELD accepted by ICLR 2026
Dec. 2025 Passed the Ph.D. Qualifying Exam, thanks to my advisor and committee members.
Aug. 2025 One paper D-CoDe accepted by EMNLP 2025 May 2025 Started Research Internship at Adobe Research. Sep. 2024 Started my journey at Northeastern University. Experience
SMILE Lab, Northeastern University, Boston
Ph.D. Student, Sep. 2024 – Present
Supervisor: Prof. Yun Raymond Fu
Adobe Research, San Jose
Research Intern (Return), May 2026 – Present
Mentor: Zhaowen Wang; Shraman Pramanick
Adobe Research, San Jose
Research Intern, May 2025 – Nov. 2025
Xidian University, Xi'an
Master Student, Sep. 2021 – Jun. 2024
Undergraduate Student, Sep. 2017 – Jun. 2021
Supervisor: Prof. Xuefeng Liang
Publications [Google Scholar]
Under Review
StreamVLM: Adapting Offline Vision-Language Models for Streaming Understanding
TL;DR: Adapts offline VLMs to streaming video understanding with event-based KV-cache memory construction and KNN-based retrieval.
Under Review
From LLM-Based World Knowledge to Physical AI: A Survey and Roadmap
TL;DR: Surveys Physical AI through LLM-based world knowledge, connecting multimodal grounding, action grounding, world modeling, policy learning, and embodied deployment.
Under Review
Selective Reduction in Foundation Models: A Survey
TL;DR: Unifies input-side, model-internal, and inference-time selection methods as a performance-oriented principle for foundation models.
Under Review
Rethinking Fine-Tuning: Unlocking Hidden Capabilities in Vision-Language Models
TL;DR: Applied MFT to VLMs: learnable gating reorganizes subnetworks without weight updates; outperforms LoRA and full fine-tuning.
ECCV 2026
Beyond Atomic Layouts: Compositional Design Understanding with Vision-Language Models
TL;DR: Diagnosed failure modes in layered designs (semantic drift, structural ambiguity) and built MASON, a plug-and-play framework with metadata-aware alignment and structural cue injection.
ACL 2026
Distorted or Fabricated? A Survey on Hallucination in Video LLMs
TL;DR: Authored a survey on Video-LLM hallucinations (taxonomy, benchmarks, mitigations) and maintain a curated repo.
IEEE FG 2026
IDEA: Capturing Individual Differences of Facial Expression for Authentic Expression Generation
TL;DR: Trained a CLIP-style contrastive model with identity and expression modalities to guide text-to-image generation toward more authentic, individual-specific expressions.
ICLR 2026
SHIELD: Suppressing Hallucinations In LVLM Encoders via Bias and Vulnerability Defense
TL;DR: Identified encoder-side causes of hallucinations and developed SHIELD, a training-free token-editing module (re-weighting + adversarial decoding) for captioning and VQA.
EMNLP 2025
D-CoDe: Scaling Image-Pretrained VLMs to Video via Dynamic Compression and Question Decomposition
TL;DR: Developed D-CoDe, a plug-and-play pipeline with dynamic compression and question decomposition for long-video QA under tight context.
ICASSP 2025
LipReading for Low-resource Languages by Language Dynamic LoRA
TL;DR: Developed dynamic LoRA for meta lip shapes and multilingual instruction tuning to improve cross-lingual lipreading in low-resource settings.
ACMMM 2021
CALLip: Lipreading using Contrastive and Attribute Learning
TL;DR: Proposed CALLip, leveraging attribute learning to normalize cross-speaker variation and audio-visual contrastive learning to mitigate viseme confusion.
Academic Service
Conference Reviewer NeurIPS, FG, ARR
Journal Reviewer ACM TKDD
Honors & Awards
2022 Outstanding Student, Xidian University
2021 National Scholarship, China
2021 Undergraduate Computer Design Competition (1st Prize), China
2019 RoboMaster National Robotics Competition (2nd Prize), China
2019 ICRA AI Challenge (3rd Prize)
Teaching Experience
Fall 2025 TA — DS 5110 Essentials of Data Science
Spr. 2026 TA — DS 5020 Fundamentals of Linear Algebra and Probability
Contact
WeChat hukcc369