Shengzhong Liu
Tenured Track Associate Professor
Works on edge intelligence, multimodal sensing and real-time scheduling for AI workloads.
Shengzhong Liu's lab at SJTU works on distributed, cloud & edge systems, vision-language & multimodal and rendering & appearance. 39 papers since 2023, 18 at top venues, mostly ICML and KDD. Applications close .
At a glance
- papers, 2023–2026
- 10/ year
- at T1 venues
- 46%
- PhD students
- —
- MS students
- —
- alumni listed
- —
Papers by yearfrom the lab homepage, 2023–2026
- 13 papers2026 · T1 9 · collecting
- 11 papers2025 · T1 4
- 9 papers2024 · T1 2
- 6 papers2023 · T1 3
T1 venuesOtherVenues: ICML, KDD, SenSys, WWW, NeurIPS
Research focuspaper titles that use each topic's words
- Distributed, cloud & edge systems4 papers
- Vision-language & multimodal4 papers
- Rendering & appearance1 paper
- multimodal sensing0 papers
- edge intelligence0 papers
Paper list
From the lab homepage.
202613 papers
- KDD (T1)$C^2KV$: Compressed and Composable KV Cache Reuse for Efficient LLM Inference
- Haoyu ZhaoBridging NL2SQL for CAN Signal Analytics at NIO: From Externalized Schemas to CTE Pipelines
- ICML (T1)BubbleSpec: Turning Long-Tail Bubbles into Speculative Rollout Drafts for Synchronous Reinforcement Learning
- CVPR (T1)CaT-GS: Efficient 3DGS Rendering for Large Scale Scenes via Inter-frame Caching and Tile Scheduling
- —CoSR: Real-Time Super-Resolution for Mobile Video Conference via Inter-Processor Collaboration
- ICML (T1)Dywave: Event-Aligned Dynamic Tokenization for Heterogeneous IoT Sensing Signals
- SIGIR (T1)Learning Noise-Resilient and Transferable Graph-Text Alignment via Dynamic Quality Assessment
- ACM IMWUT/UbiComp , Shanghai, China, OctoberMACRA: Efficient Knowledge Sharing among Personalized Models against Label Shift in Mobile AI Applications
- —REMIX: Adaptive Open-Domain Signal Mixing for Efficient Sensing Data Augmentation
- KDD (T1)RouteGoT: Node-Adaptive Routing for Cost-Efficient Graph of Thoughts Reasoning
- ICML (T1)SmartThinker: Progressive Chain-of-Thought Length Calibration for Efficient Large Language Model Reasoning
- EuroSys (T1)TokenFlow: Responsive LLM Text Streaming Serving under Request Burst via Preemptive Scheduling
202511 papers
- ACL (Oral), Vienna, Austria, July$Pre^3$: Enabling Deterministic Pushdown Automata for Faster Structured LLM Generation
- WWW (Oral), Sydney, Australia, AprilABO: Abandon Bayer Filter for Adaptive Edge Offloading in Responsive Augmented Reality
- NeurIPS (T1)AdaTS: Learning Adaptive Time Series Representations via Dynamic Soft Contrasts
- —AutoMOCHA: Automated Adaptation Hierarchy for Mobile Video Analytics Against Domain Shifts
- IEEE ICDCS , Glasgow, UK, JulyDistributed DNN-based Video Analytics with Adaptive Multi-Device Collaboration
- WWW (T1)Enabling Real-Time Inference in Online Continual Learning via Device-Cloud Collaboration
- WWW (T1)InfoMAE: Pairing-Efficient Cross-Modal Alignment with Informational Masked Autoencoders for IoT Signals
- IEEE RTSS , Boston, USA, DecemberNova: Real-Time Agentic Vision-Language Model Serving with Adaptive Cross-Stage Parallelization
- ACM IMWUT/UbiComp , Espoo, Finland, OctoberOpenMAE: Efficient Masked Autoencoder for Vibration Sensing with Open-Domain Data Enrichment
- SenSys (T1)Responsive DNN Adaptation for Video Analytics against Environment Shift via Hierarchical Mobile-Cloud Collaborations
- ACM IMWUT/UbiComp , Espoo, Finland, OctoberSemiCMT: Contrastive Cross-Modal Knowledge Transfer for IoT Sensing with Semi-Paired Multi-Modal Signals
20249 papers
- —Algorithms for Canvas-based Attention Scheduling with Resizing
- IEEE RTSS, York, UK, DecemberFLEX: Adaptive Task Batch Scheduling with Elastic Fusion in Multi-Modal Multi-View Machine Perception
- WWW (T1)FreqMAE: Frequency-Aware Masked Autoencoder for Multi-Modal IoT Sensing
- —MetaHKG: Meta Hyperbolic Learning for Few-shot Temporal Reasoning
- FM-Sys Workshop (at CPS-IoT Week), Hong Kong, China, MayOn the Efficiency and Robustness of Vibration-based Foundation Models for IoT Sensing: A Case Study
- SenSys (T1)PhyMask: An Adaptive Masking Paradigm for Efficient Self-Supervised Learning in IoT
- —Taming Algorithmic Priority Inversion in Mission-critical Perception Pipelines
- —TGOnline: Enhancing Temporal Graph Learning with Adaptive Online Meta-Learning
- —VibroFM: Towards Micro Foundation Models for Robust Multimodal IoT Sensing
20236 papers
- NeurIPS (T1)FOCAL: Contrastive Learning for Multimodal Time-Series Sensing Signals in Factorized Orthogonal Latent Space
- —Generalized Self-Cueing Real-Time Attention Scheduling with Intermittent Inspection and Image Resizing
- ACL (T1)Noisy Positive-Unlabeled Learning with Self-Training for Speculative Knowledge Graph Reasoning
- SenSys (T1)SudokuSens: Enhancing Deep Learning Robustness for IoT Sensing Applications using a Generative Approach
- IEEE ICCCN, Honolulu, USA, JulyTwinSync: A Digital Twin Synchronization Protocol for Bandwidth-limited IoT Applications
- IEEE ICCCN, Honolulu, USA, JulyUnderprovisioned GPUs: On Sufficient Capacity for Real-Time Mission-Critical Perception
Where alumni went
Alumni are not listed on the lab homepage.
Apply to SJTUFall 2027
Computer Science PhD (international applicants)
- Round
- Self-funded
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- GBCambridge
Nicholas Lane
- Efficient training & inference
- GBUCL
Paolo Barucca
- Computational social science
- USMIT
Devavrat Shah
- Learning theory & optimization
- GBUCL
Maria Del Rio-Chanona
- Large language models
- CNSJTU
Luoyi Fu
- Graph mining & networks
- CNPeking
Hong Mei
- Graph learning
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.
SJTUcloses