Xinbao Qiao
Latest Updates8 updatesRecognized as a NeurIPS 2026 Top ReviewerRecognition for contributions to peer review.
- Recognized as a NeurIPS 2026 Top Reviewer
Recognition for contributions to peer review.
- Started PhD studies at CUHK
PhD in Information Engineering, The Chinese University of Hong Kong.
- Completed master’s degree
M.Eng. in Artificial Intelligence, Zhejiang University.
- Started full-time research internship
Six-month full-time research internship on trustworthy LLMs at NUSRI-CQ.
Site activityAll history
Browse Xinbaopedia
Browse view
Research topics
Core research
- ResearchXinbao Qiao's research on data-centric machine learning, AI and networks, and trustworthy AI.
- AI and NetworksPrimary research topic for Qiao Xinbao, covering AI systems under networked data and communication constraints.
- Data Centric MLResearch topic focused on data quality, selection, valuation, correction, and governance.
Methods and geometry
- Machine UnlearningResearch topic on removing or correcting data influence from trained models.
- Distributed Wasserstein BarycenterComputing a shared Wasserstein reference distribution from locally held data.
- Influence FunctionsMethods for estimating how training examples affect learned models.
Reliability and trust
- Synthetic DataResearch topic on synthetic data, recursive training, low-resource verification, selection bias, and model collapse.
- Fairness and RobustnessFairness and robustness as goals of data-driven model correction.
- LLM ReliabilityReliability of large language model systems in realistic use.
Publications and projects
Indexes
- PublicationsPublication list for Qiao Xinbao.
- ProjectsResearch projects in AI and networks, machine unlearning, synthetic data, and reliable language models.
Selected publications
- Illusory Pattern Perception Drives Spurious Inference in Large Language ModelsNeurIPS 2026 paper comparing LLM and human illusory pattern perception across three psychological tasks, with sparse autoencoder feature analysis.
- Hessian-Free Online Certified UnlearningICLR 2025 paper on efficient Hessian-free certified machine unlearning.
- Beyond Binary Erasure: Soft-Weighted Unlearning for Fairness and RobustnessAAAI 2026 paper on soft-weighted unlearning for fairness and robustness correction.
- When Sample Selection Bias Precipitates Model CollapseICML 2026 paper on low-resource verification regimes, sample-selection bias, model collapse, and collaborative Wasserstein-geometry proxies.
- DynFrs: An Efficient Framework for Machine Unlearning in Random ForestICLR 2025 paper on efficient machine unlearning for random forests.
- Decentralized Free-Support Wasserstein BarycenterUnder-review manuscript on movable barycenter supports and shared geometric references under decentralized communication.
Project pages
- AI and NetworksPrimary research topic for Qiao Xinbao, covering AI systems under networked data and communication constraints.
- Collaborative EvaluationEvaluation of models and data processes using evidence from multiple parties.
Affiliations
Profile
- Curriculum VitaeAcademic CV summary for Xinbao Qiao.
- EducationEducation timeline for Qiao Xinbao.
- ExperienceResearch experience and affiliations.
Institutions
- The Chinese University of Hong KongCurrent doctoral institution of Xinbao Qiao.
- Zhejiang UniversityMaster's institution of Xinbao Qiao.
- Shandong UniversityUndergraduate institution of Xinbao Qiao.
- NUSRI-CQResearch internship institution of Xinbao Qiao.
Academic network
- Angela Yingjun ZhangCUHK Information Engineering professor and doctoral advisor of Xinbao Qiao.
- Meng ZhangZhejiang University faculty member and master's advisor of Xinbao Qiao.