Hongsheng Hu
Tenured Track Associate Professor
Works on trustworthy machine learning, focusing on privacy vulnerabilities such as membership inference and their defenses.
Hongsheng Hu's lab at SJTU works on trustworthy & safe ML, efficient training & inference and interpretability, evaluation & safety of LMs. 33 papers since 2023, 16 at top venues, mostly IJCAI and AAAI. Applications close .
At a glance
- papers, 2023–2026
- 9/ year
- at T1 venues
- 48%
- PhD students
- —
- MS students
- —
- alumni listed
- —
Papers by yearfrom DBLP, 2023–2026
- 8 papers2026 · T1 3 · collecting
- 11 papers2025 · T1 6
- 11 papers2024 · T1 6
- 3 papers2023 · T1 1
T1 venuesOtherVenues: IJCAI, AAAI, NDSS, NeurIPS, WWW
Research focuspaper titles that use each topic's words
- Efficient training & inference8 papers
- membership3 papers
- membership inference3 papers
- Trustworthy & safe ML0 papers
- Interpretability, evaluation & safety of LMs0 papers
Paper list
From DBLP.
20268 papers
- IEEE Trans. Inf. Forensics Secur.DP2-RAG: An Efficient Full-Process Differential Privacy Implementation in Retrieval-Augmented Generation (opens in a new tab)
- NDSS (T1)ExpShield: Safeguarding Web Text from Unauthorized Crawling and LLM Exploitation (opens in a new tab)
- IEEE Trans. Dependable Secur. Comput.From Pixels to Trajectory: Universal Adversarial Example Detection via Temporal Imprints (opens in a new tab)
- Des. Codes Cryptogr.Infinite families of constacyclic codes supporting 3-designs (opens in a new tab)
- Eng. Appl. Artif. Intell.Intellectual property protection for deep learning model and dataset intelligence (opens in a new tab)
- AAAI (T1)Reference Recommendation Based Membership Inference Attack Against Hybrid-Based Recommender Systems (opens in a new tab)
- SPSoK: Robustness in Large Language Models against Jailbreak Attacks (opens in a new tab)
- ICLR (T1)Unlearning during Training: Domain-Specific Gradient Ascent for Domain Generalization (opens in a new tab)
202511 papers
- IEEE Trans. Ind. Electron.A High Power Density and Nearly Constant Voltage WPT System Based on Bendable PCB (opens in a new tab)
- Sci. China Inf. Sci.Artificial intelligence security and privacy: a survey (opens in a new tab)
- RAIDBadFU: Backdoor Federated Learning through Adversarial Machine Unlearning (opens in a new tab)
- IJCAI (T1)Balancing Invariant and Specific Knowledge for Domain Generalization with Online Knowledge Distillation (opens in a new tab)
- ACM MMBoosting Guided Diffusion with Large Language Models for Multimodal Sequential Recommendation (opens in a new tab)
- CVPR (T1)Enhancing Adversarial Transferability with Checkpoints of a Single Model's Training (opens in a new tab)
- IJCAI (T1)Fine-Grained and Efficient Self-Unlearning with Layered Iteration (opens in a new tab)
- NeurIPS (T1)HPSERec: A Hierarchical Partitioning and Stepwise Enhancement Framework for Long-tailed Sequential Recommendation (opens in a new tab)
- WWW (T1)MGF-ESE: An Enhanced Semantic Extractor with Multi-Granularity Feature Fusion for Code Summarization (opens in a new tab)
- AsiaCCSWhen Better Features Mean Greater Risks: The Performance-Privacy Trade-Off in Contrastive Learning (opens in a new tab)
- IJCAI (T1)Where Does This Data Come From? Enhanced Source Inference Attacks in Federated Learning (opens in a new tab)
202411 papers
- NDSS (T1)A Duty to Forget, a Right to be Assured? Exposing Vulnerabilities in Machine Unlearning Services (opens in a new tab)
- WWW (T1)Cardinality Counting in "Alcatraz": A Privacy-aware Federated Learning Approach (opens in a new tab)
- NeurIPS (T1)DapperFL: Domain Adaptive Federated Learning with Model Fusion Pruning for Edge Devices (opens in a new tab)
- USENIX Security (T1)DNN-GP: Diagnosing and Mitigating Model's Faults Using Latent Concepts (opens in a new tab)
- J. Frankl. Inst.H∞ security control for 2-D switched time delay systems under arbitrary delay time-variation rate and stochastic cyber attacks (opens in a new tab)
- SPLACMUS: Latent Concept Masking for General Robustness Enhancement of DNNs (opens in a new tab)
- SPLearn What You Want to Unlearn: Unlearning Inversion Attacks against Machine Unlearning (opens in a new tab)
- IEEE Trans. Ind. Electron.Modularized Inductive Power Transfer Systems With Inherent Impedance Decoupling for High-Power Applications (opens in a new tab)
- IJCAI (T1)Shadow-Free Membership Inference Attacks: Recommender Systems Are More Vulnerable Than You Thought (opens in a new tab)
- IEEE Trans. Dependable Secur. Comput.Source Inference Attacks: Beyond Membership Inference Attacks in Federated Learning (opens in a new tab)
- AAAI (T1)Symmetric Self-Paced Learning for Domain Generalization (opens in a new tab)
20233 papers
- Concurr. Comput. Pract. Exp.Differentially private locality sensitive hashing based federated recommender system (opens in a new tab)
- IJCAI (T1)OptIForest: Optimal Isolation Forest for Anomaly Detection (opens in a new tab)
- AsiaCCSPOSTER: ML-Compass: A Comprehensive Assessment Framework for Machine Learning Models (opens in a new tab)
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
- USUC Berkeley
Nika Haghtalab
- Learning theory & optimization
- USNYU
Rudy Morel
- Learning theory & optimization
- SGNTU
Tianwei Zhang
- Trustworthy & safe ML
- GBImperial
Stefan Vlaski
- Learning theory & optimization
- USUCLA
Guy Van den Broeck
- Probabilistic & causal inference
- USCMU
Vincent Conitzer
- Reasoning & neuro-symbolic AI
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
- DBLP
- Members
- Not collected
- Alumni
- Not collected
Tiers are venue classifications, not lab rankings.
SJTUcloses