Gongyou Xu (许恭友)

Gongyou Xu speaking at a research presentation

About

I am a Research Fellow at Beijing Normal–Hong Kong Baptist University (BNBU). My work lies at the intersection of graph learning, recommender systems, multimodal representation learning, and efficient AI systems.

I have contributed to peer-reviewed and under-review research while also developing real-world systems for vessel safety, indoor positioning, knowledge services, document intelligence, and distributed LLM inference. I am interested in methods that are technically rigorous, reproducible, and useful in practice.

My current research interests include:

  • Graph machine learning and graph condensation.
  • Sequential recommendation and efficient learning objectives.
  • Multimodal representation learning and neural topic modeling.
  • Efficient and distributed large language model inference.

I am currently looking for opportunities to pursue graduate or doctoral study, and I would be grateful for the chance to contribute to academic collaborations.

Publications

My name is shown in bold.

Simplified Sampled Softmax Loss for Sequential Recommendation

Gongyou Xu, A. Zhang, Y. Yu, L. Zhang, R. Gao, J. Wei, H. Yin

Submitted to COLING 2026

GCRank: Graph Condensation for Link Prediction via Edge Rank Preservation

W. Zhu, Gongyou Xu, Z. Yu, S. Wang, A. C. Zhou, J. Chen, et al.

Manuscript in preparation for submission to ICLR 2027

Transporting Semantics Beyond Words: Multimodal Augmented Embedding Transport for Neural Topic Modeling

D. Guo, Z. Luo, Gongyou Xu, N. Bouguila, X. Liu, W. Fan

Submitted to AAAI 2027

PASTD: Progressive Augmentation and Spatiotemporal Decoupling Contrastive Learning for Skeleton-Based Action Recognition

Q. Huang, W. Qian, C. Li, Gongyou Xu, Z. Chen

IEEE ICASSP, 2025

Hyperbolic Adversarial Learning for Personalized Item Recommendation

A. Zhang, Y. Yu, Gongyou Xu, R. Gao, L. Zhang, S. Gao, H. Yin

DASFAA, 2024

Design and Deployment of Django-based Housing Information Management System

X. Yu, X. Li, C. Wu, Gongyou Xu

Journal of Physics: Conference Series, 2023

Research Experience

Beijing Normal–Hong Kong Baptist University

Apr. 2024 – Present

Research Fellow

  • Research on graph learning, sequential recommendation, multimodal topic modeling, and distributed LLM inference.
  • Model implementation, data preparation, experimental evaluation, and academic writing for collaborative research projects.
  • Developed reproducible Docker environments and a graph-compression method retaining about 95% of baseline performance at around 80% compression.

Fujian Cross-Strait Information Technology Co., Ltd.

Apr. 2023 – Apr. 2024

AI / Large Language Model Engineer

  • Developed applied AI systems across knowledge services, document intelligence, computer vision, indoor positioning, and workflow automation.
  • Worked with PyTorch, OCR, YOLO, Django, REST APIs, Docker, Linux, WebSocket, MQTT, and data pipelines.

Selected Projects

Distributed LLM Inference System Based on vLLM 2026 – Present

Designed a multi-node inference workflow covering service decomposition, node communication, transport optimization, containerized deployment, and concurrency testing.

Vision-Based Vessel Personnel Safety System 2023 – 2024

Developed person detection, behavior recognition, and fall-overboard alerts from surveillance video, with MQTT-based on-site integration.

High-Precision UWB Indoor Positioning System 2023 – 2024

Worked on positioning algorithms, data communication, and real-time WebSocket links for a major power-grid enterprise.

Complex Layout OCR Platform 2023 – 2024

Developed preprocessing, layout analysis, recognition, and extraction workflows for historical books and mixed-layout documents.

National Genealogy and Cultural Knowledge System 2023 – 2024

Contributed retrieval, entity and relation extraction, association analysis, and question answering; the system was demonstrated at the Straits Forum.

Education

Nanjing University of Posts and Telecommunications, Tongda College

2019 – 2023

B.Eng. in Computer Science and Technology

Undergraduate thesis: oriented object detection for optical remote-sensing imagery.

Awards