2026年软件理论讨论班日程安排

注1:2026年春季学期讨论班时间未特别说明均为周三上午10:00-12:00。

注2:有意在讨论班做学术报告的校外学者请直接email联系:sunmeng@math.pku.edu.cn。 

 

时间(Date)

地点(Place)

报告题目(Title)

报告人 (Speaker)

1
2026.3.4
知无涯-313
关于科研和论文写作的一些建议

孙猛

2
2026.3.11
知无涯-313
From Monolithic to Compositional: A Compositional Operational Semantics for Crystality
徐紫云
3
2026.3.18
知无涯-313
ClawWorm: Self-Propagating Attacks Across LLM Agent Ecosystems
张益豪
4
2026.3.25
知无涯-313
利用蜕变关系进行强化学习蒸馏
邵凯诚
5
2026.4.1
知无涯-313
PROBIX: Probing-Guided Suffix Attacks on Large Language Models
林与心
6
2026.4.8
知无涯-313

Monitoring LLM-based Multi-Agent Systems Against Corruptions via Node Evaluation

A Closed-Loop Framework for Testing, Interpreting, and Optimizing Reinforcement Learning based on Metamorphic Relations

邬程灿

邵凯诚

7
2026.4.15
知无涯-313
Beyond Translation: A Decomposed Collaborative Reasoning Framework Harnessing LLMs and Symbolic Solvers
张知辛
8
2026.4.22
知无涯-313

综合考试

开题报告

林与心

邵凯诚、张益豪

9
2026.4.29
知无涯-313
Denotational Semantics of Crystality
徐紫云
10
2026.5.6
知无涯-313
FM-Agent: Scaling Formal Methods to Large Systems via LLM-Based Hoare-Style Reasoning
丁浩然(MIT)
11
2026.5.13
知无涯-313
A Survey of Policy Distillation: From Fundamental to Frontier
邵凯诚
12
2026.5.20
知无涯-313
AutoML for Spatio-temporal Data: from Raw data to Decisions
Mitra Baratchi(Leiden University)
13
2026.5.27
知无涯-313

深度神经网络的知识产权保护与输出规制

建模语言MedTiny的设计及其编译器实现

栾晓坤

李翔宇

14
2026.6.3
知无涯-313

Formal Runtime Verification for File-Manipulating Code Agents

SoK: System Prompt Leakage in LLM Agents Reconnaissance, Consequence, and the Architectural Escape

张益豪
15
2026.6.10
知无涯-313

ReGA: Model-Based Safeguard for LLMs via Representation-Guided Abstraction

魏泽明

往年讨论班信息

2025

2024

2023

2022

2021

2019

2018

2017

2016

2015