Researcher — paper.dinq.me

作者位次 #5/7ICML 2026Poster· Deep Learning / Generative Models and Autoencoders

When Sample Selection Bias Precipitates Model Collapse

Xinbao Qiao · Xianglong Du · Wei Liu · Jingqi Zhang · Peihua Mai · Meng Zhang · Yan (James) Pang

一作ACL 2026Poster· Natural Language Processing

S hared R equest: Privacy-Preserving Model-Agnostic Inference for Large Language Models

Peihua Mai · Xuanrong Gao · Youlong Ding · Xianglong Du · Wei Liu · Yan Pang

摘要

With the widespread deployment of public large language models (LLMs) such as ChatGPT, protecting user prompt privacy has become an increasingly critical issue. Existing privacy-preserving inference methods sacrifice either utility or efficiency, and often require model-specific modifications tha...

效率与系统安全可解释性OpenReview 项目页

Peihua Mai

获取完整研究档案

联系方式

2026 顶会论文

When Sample Selection Bias Precipitates Model Collapse

S hared R equest: Privacy-Preserving Model-Agnostic Inference for Large Language Models

常见合作者