报告题目：A Statistical Hypothesis Testing Framework for Data Mi…

报 告 人：张林俊

时 间：6月20日（周五）上午10：00 - 11：00

地 点：新文科楼403会议室

简 介：Large Language Models (LLMs) are rapidly gaining enormous popularity in recent years. However, the training of LLMs has raised significant privacy and legal concerns, particularly regarding the inclusion of copyrighted materials in their training data without proper attribution or licensing, which falls under the broader issue of data misappropriation. In this article, we focus on a specific problem of data misappropriation detection, namely, to determine whether a given LLM has incorporated data generated by another LLM. To address this issue, we propose embedding watermarks into the copyrighted training data and formulating the detection of data misappropriation as a hypothesis testing problem. We develop a general statistical testing framework, construct a pivotal statistic, determine the optimal rejection threshold, and explicitly control the type I and type II errors. Furthermore, we establish the asymptotic optimality properties of the proposed tests, and demonstrate its empir...

报告题目：人工智能与定理证明

报 告 人：王善文

时 间：2025年6月19日（周四）晚上19:00-20:00

地 点：国交2号楼315会议室

简 介：在本报告中, 我们将从数学研究的主要逻辑范式来分析当前人工智能在定理证明方面的进展, 并介绍交互式定理证明助手Lean.

报告题目：Continuous Modeling Perspective for Imaging Science

报 告 人：赵熙乐

时 间：2025年6月23日（周一）上午10:00--11:00

地 点：国交2号楼201

简 介：To tackle inverse problems in imaging science, the regularizer, serving as an indispensable cornerstone in modeling, are usually introduced. In this talk, we will begin by reviewing the classical regularizers, including local regularizers, nonlocal regularizers, and global regularizers. We then will discuss the limitations of classical hand-crafted regularizers (e.g., expressive capability, applicability, and flexibility). To address the above limitations of classical regularizers, we suggest a unified Continuous Modeling Perspective for imaging science,  which continuously represents discrete data by elegantly leveraging tiny neural networks. This paradigm allows us to readily deconstruct and reconstruct the classical regularizers, thus  unleashing the potential of regularizers. Extensive experiments demonstrate the promising performance of the continuous modeling perspective.