DONGGUK UNIVERSITY
Intelligent Systems
& Networking Lab.
동국대학교 시스템 소프트웨어 및 네트워크 연구실
We explore the synergy between system and network for AI, and AI for system and network. Our research targets faster, more efficient, and more reliable computing infrastructures from edge devices to cloud and datacenter-scale systems.
Recruiting We are looking for highly motivated undergraduate and graduate students! AI 시스템, 클라우드, 네트워크 연구에 관심이 있는 열정 많은 학생의 합류를 기대합니다!News
more- 🎉 One paper got accepted at Scientific Reports (SCIE, Q1). Efficient Training-free Query Routing Framework for Federated Retrieval-Augmented Generation
- 👨💻 Four interns have officially begun their master courses. We sincerely hope they have a meaningful and rewarding journey ahead!
- 🎉 One paper got accepted at EMNLP 2026 (main) (Top CS conference, BK 3). Residual Sparsification via Output Importance for Compressing Mixture-of-Experts LLMs
- 🎓 Six undergraduate interns received their B.S. degrees. Congratulations!
- 🎉 Two papers got accepted at IEEE MASCOTS 2026 (Top CS conference, BK 2). Accurate Simulation of Distributed Training Jobs with Network Contention Modeling Xronos: Heterogeneity-Aware Tensor Parallelism for Collaborative LLM Fine-Tuning on Edge CPUs
- 🏆 An undergraduate capstone design team advised by Prof. Yeonho Yoo received the Paper Award (장려상) at KCC 2026.
- 🎉 One paper got accepted at Future Generation Computer Systems Journal (Top 10% Journal). Prediction-based GPU Sharing for Distributed Training
Research
Explore ResearchOur research theme is System for AI — building the system software and networking infrastructure that AI actually runs on, from GPU clusters and datacenter fabrics down to edge devices and IoT.
Selected Publications
Browse Publications- Residual Sparsification via Output Importance for Compressing Mixture-of-Experts LLMs EMNLP 2026 (main) DOI
- Accurate Simulation of Distributed Training Jobs with Network Contention Modeling IEEE MASCOTS 2026 DOI
- Xronos: Heterogeneity-Aware Tensor Parallelism for Collaborative LLM Fine-Tuning on Edge CPUs IEEE MASCOTS 2026 DOI
- Revisiting Traffic Splitting for Software Switch in Datacenter ACM SIGMETRICS 2025 DOI
- Prediction-based GPU Sharing for Distributed Training Future Generation Computer Systems Journal DOI
- Parameter-Efficient 12-Lead ECG Reconstruction from a Single Lead MICCAI 2025 DOI
- Intelligent Packet Processing for Performant Containers in IoT IEEE Internet of Things Journal DOI
- Machine Learning-Based Prediction Models for Control Traffic in SDN Systems IEEE Transactions on Services Computing DOI
Contact
Students interested in system software, networking, cloud/edge computing, distributed AI training, and AI-driven optimization are welcome to contact the lab.
View full contact & join us →Dept. of Computer Science and Artificial Intelligence, Dongguk University
Office: Bldg. P #621 · Lab: Bldg. P #P636
30, Pildong-ro 1-gil, Jung-gu, Seoul 04620