Kai Chen
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
Research Interests
Kai Chen has a total of 406 co-authors. Here are the top co-authors by number of shared publications:
- Junxue Zhang
University of Science and Technology of China, Hefei, China
26 shared publications - Qiang Yang
The Hong Kong Polytechnic University, Hong Kong, Hong Kong
18 shared publications - Han Tian
University of Science and Technology of China, Hefei, China
16 shared publications - Wenxue Li
Hong Kong University of Science and Technology, Hong Kong, Hong Kong
16 shared publications Yan Chen
Northwestern University
15 shared publications- Di Chai
Shanghai University of Finance and Economics, Shanghai, China
14 shared publications
Latest Publications
- Open Access
- 17 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 14-26
FCP: Fast User-space RDMA Control Plane Setup for Elastic Computing
- 17 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 14-26
Open AccessFCP: Fast User-space RDMA Control Plane Setup for Elastic Computing
- Open Access
- 14 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 1341-1355
PrefixFlow: Training-Time KV Caching for Schedule-Level Prefix Reuse in LLM RL Training
- 14 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 1341-1355
Open AccessPrefixFlow: Training-Time KV Caching for Schedule-Level Prefix Reuse in LLM RL Training
- Open Access
- 14 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 597-611
D-Pingmesh: Dynamic Probing and Fault Localization for Reconfigurable Datacenter Networks
- 14 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 597-611
Open AccessD-Pingmesh: Dynamic Probing and Fault Localization for Reconfigurable Datacenter Networks
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (11) , 7492-7503
Enhancing LLMs for Expert Question Answering by Synergizing With Knowledge Graphs
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (11) , 7492-7503
Enhancing LLMs for Expert Question Answering by Synergizing With Knowledge Graphs
- 6 September 2026
- Published by Elsevier in Information Processing & Management
- Vol. 64 (2)
Attention consistency matters: Self-consistent learning with counterfactual regularization for street-level IP geolocation
- 6 September 2026
- Published by Elsevier in Information Processing & Management
- Vol. 64 (2)
Attention consistency matters: Self-consistent learning with counterfactual regularization for street-level IP geolocation
- Open Access
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 60-73
TurboBus: Pooling PCIe Bandwidth for LLM Workloads via Scale-Up Fabrics
GPU memory offloading is widely adopted for LLM workloads but shifts the bottleneck to GPU-CPU transfers, which can take up to 90% of the end-to-end inference/training time! Paradoxically, over 60% of PCIe bandwidth remains idle. The root cause is that PCIe links are individually bottlenecked but collectively underutilized. Bursty, phase-driven transfer patterns leave bandwidth idle both within and across jobs. We present TurboBus, which pools PCIe bandwidth across co-located jobs via emerging scale-up fabrics. TurboBus enables any GPU to borrow idle PCIe links from neighboring GPUs, even those belonging to other jobs, while preserving isolation through a privileged daemon. At the core of TurboBus, it streams data through relay GPUs with bounded memory overhead, keeps all links busy through fine-grained PCIe allocation, enables bidirectional transfers leveraging PCIe/NVLink bandwidth asymmetry, and balances fairness and completion time with a size-aware scheduling. We fully implement TurboBus and our experiments show that it reduces first-token latency by up to 40% for on-demand model loading (within 5% of the analytical optimum), achieves up to 1.6x throughput for KV-cache-offloaded inference, and accelerates training by up to 7%, while imposing less than 1% overhead on co-located workloads.
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 60-73
Open AccessTurboBus: Pooling PCIe Bandwidth for LLM Workloads via Scale-Up Fabrics
GPU memory offloading is widely adopted for LLM workloads but shifts the bottleneck to GPU-CPU transfers, which can take up to 90% of the end-to-end inference/training time! Paradoxically, over 60% of PCIe bandwidth remains idle. The root cause is that PCIe links are individually bottlenecked but collectively underutilized. Bursty, phase-driven transfer patterns leave bandwidth idle both within and across jobs. We present TurboBus, which pools PCIe bandwidth across co-located jobs via emerging scale-up fabrics. TurboBus enables any GPU to borrow idle PCIe links from neighboring GPUs, even those belonging to other jobs, while preserving isolation through a privileged daemon. At the core of TurboBus, it streams data through relay GPUs with bounded memory overhead, keeps all links busy through fine-grained PCIe allocation, enables bidirectional transfers leveraging PCIe/NVLink bandwidth asymmetry, and balances fairness and completion time with a size-aware scheduling. We fully implement TurboBus and our experiments show that it reduces first-token latency by up to 40% for on-demand model loading (within 5% of the analytical optimum), achieves up to 1.6x throughput for KV-cache-offloaded inference, and accelerates training by up to 7%, while imposing less than 1% overhead on co-located workloads.
- 1 August 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10 (4) , 3207-3220
A Federated Split Autoencoder for Privacy-Preserving Non-IID Graph Learning
- 1 August 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10 (4) , 3207-3220
A Federated Split Autoencoder for Privacy-Preserving Non-IID Graph Learning
- 16 July 2026
- Published by Elsevier in Computer Networks
- Vol. 287
AdaGeo: Post-deployment IP geolocation via source-free test-time adaptation
- 16 July 2026
- Published by Elsevier in Computer Networks
- Vol. 287
AdaGeo: Post-deployment IP geolocation via source-free test-time adaptation
- 30 June 2026
- Published by Springer Nature
- p. 411-422
Federated Co-tuning Framework for Large and Small Language Models
- 30 June 2026
- Published by Springer Nature
- p. 411-422
Federated Co-tuning Framework for Large and Small Language Models