Xupeng Miao
Assistant Professor
Researches machine learning systems, data management and distributed computing for training and serving large models.
Xupeng Miao's lab at Peking works on efficient training & inference, distributed, cloud & edge systems and data management & big data. 44 papers since 2023, 6 at top venues, mostly VLDB and TKDE. Applications close .
At a glance
- papers, 2023–2026
- 11/ year
- at T1 venues
- 14%
- PhD students
- —
- MS students
- —
- alumni listed
- —
Papers by yearfrom the lab homepage, 2023–2026
- 12 papers2026 · T1 0 · collecting
- 8 papers2025 · T1 1
- 16 papers2024 · T1 2
- 8 papers2023 · T1 3
T1 venuesOtherVenues: VLDB, TKDE, VLDB Journal
Research focuspaper titles that use each topic's words
- Efficient training & inference21 papers
- Distributed, cloud & edge systems6 papers
- Data management & big data2 papers
- serving large0 papers
- distributed computing0 papers
Paper list
From the lab homepage.
202612 papers
- Proceedings of EuroSys ConferenceAdaServe: Accelerating Multi-SLO LLM Serving with SLO-Customized Speculative Decoding
- Proceedings of ICML ConferenceDARTS: Distribution-Aware Active Rollout Trajectory Shaping for Accelerating LLM Reinforcement Learning
- Proceedings of ICDE ConferenceDistVec: Efficient Distributed Machine Learning in Parallel Database Systems
- Proceedings of ICML ConferenceEchoAttention: Exploiting Token-Pair Redundancy and Frame-Block Similarity for Efficient Video Generation
- Proceedings of MLSys ConferenceEvent Tensor: A Unified Abstraction for Compiling Dynamic Megakernel
- Proceedings of NSDI ConferenceFlexLLM: Token-Level Co-Serving of LLM Inference and Fine-Tuning with SLO Guarantees
- Proceedings of OSDI ConferenceHetu v2: A General and Scalable Deep Learning System with Hierarchical and Heterogeneous Single Program Multiple Data Annotations
- Proceedings of OSDI ConferenceMirage Persistent Kernel: A Compiler and Runtime for Mega-Kernelizing Tensor Programs
- Proceedings of SOSP ConferenceMulti-LLM Serving at Production Scale
- Proceedings of NeurIPS ConferencePolySplat: Workload-Regime-Aware Rasterization for 3D Gaussian Splatting
- Proceedings of COLM ConferenceQLPO: Quadrant-weighted Sampling for Length-aware Policy Optimization
- ACM Computing SurveysTowards Efficient Generative Large Language Model Serving: A Survey from Algorithms to Systems
20258 papers
- VLDB Journal (T1)(Special Issue on Best Papers of VLDB 2022)
- Proceedings of ICML ConferenceDemystifying Cost-Efficiency in LLM Serving over Heterogeneous GPUs
- Proceedings of ASPLOS ConferenceGraphPipe: Improving Performance and Scalability of DNN Training with Graph Pipeline Parallelism
- Proceedings of ASPLOS ConferenceHelix: Distributed Serving of Large Language Models via Max-Flow on Heterogeneous GPUs
- Proceedings of OSDI ConferenceMirage: A Multi-Level Superoptimizer for Tensor Programs
- Proceedings of ICLR ConferenceNetMoE: Accelerating MoE Training through Dynamic Sample Placement
- Proceedings of SIGMOD ConferencePQCache: Product Quantization-based KVCache for Long Context LLM Inference
- Proceedings of ASPLOS ConferenceSpindle: Efficient Distributed Training of Multi-Task Large Models via Wavefront Scheduling
202416 papers
- Proceedings of ASPLOS Conference(Distinguished Artifact Award), (IEEE Micro Top Picks Honorable Mention)
- Proceedings of ACL Conference(Outstanding Paper Award)
- Proceedings of SIGMOD Conference(Tutorial)
- Proceedings of AAAI ConferenceAccelerating Text-to-image Editing via Cache-enabled Sparse Diffusion Inference
- Proceedings of SC ConferenceAtlas: Hierarchical Partitioning for Quantum Circuit Simulation on GPUs
- ACM Computing SurveysDistributed Graph Neural Network Training: A Survey
- Proceedings of SOSP ConferenceEnabling Parallelism Hot Switching for Efficient Training of Large Language Models
- VLDB (T1)Experimental Analysis of Large-scale Learnable Vector Storage Compression
- Proceedings of EACL ConferenceGenerative Dense Retrieval: Memory Can Be a Burden
- TKDE (T1)Improving Automatic Parallel Training via Balanced Memory Workload Optimization
- Proceedings of NeurIPS ConferenceLSH-MoE: Communication-efficient MoE Training via Locality-Sensitive Hashing
- Proceedings of ICDE ConferenceMFIX: An Efficient and Reliable Index Advisor via Multi-Fidelity Bayesian Optimization
20238 papers
- VLDB (T1)Angel-PTM: A Scalable and Economical Large-scale Pre-training System in Tencent
- Proceedings of AAAI ConferenceCALIP: Zero-Shot Enhancement of CLIP with Parameter-free Attention
- Proceedings of OSDI ConferenceEinNet: Optimizing Tensor Programs with Derivation-Based Transformations
- Proceedings of SIGMOD ConferenceFlexMoE: Scaling Large-scale Sparse Pre-trained Model Training via Dynamic Device Placement
- VLDB (T1)Galvatron: Efficient Transformer Training over Multiple GPUs Using Automatic Parallelism
- Proceedings of NeurIPS ConferenceModel-enhanced Vector Index
- Proceedings of IJCAI ConferenceOSDP: Optimal Sharded Data Parallel for Distributed Deep Learning
- VLDB (T1)SDPipe: A Semi-Decentralized Framework for Heterogeneity-aware Pipeline-parallel Training
Where alumni went
Alumni are not listed on the lab homepage.
Apply to PekingFall 2027
Computer Science PhD (international applicants)
- Round
- Main
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USUCSD
Sujit Dey
- Distributed, cloud & edge systems
- USUIUC
Francis Yan
- Data management & big data
- CNTsinghua
Yuan Meng
- Efficient training & inference
- USUCSD
Arun Kumar
- Data management & big data
- USUIUC
Yongjoo Park
- Efficient training & inference
- USMIT
Song Han
- Efficient training & inference
Your notes
Save this lab to keep notes.
Data
Some numbers could not be verified:
- The member list could not be found on the lab site.
- The alumni list could not be found on the lab site.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Not collected
- Alumni
- Not collected
Tiers are venue classifications, not lab rankings.
Pekingcloses