MLNLP 2026 Symposium: Top AI Scholars from Qiyuan Lab, BIT, Tsinghua & Alibaba Reveal New Agent and Table Research

MLNLP 2026 academic symposium will be held online on May 31, 2026 (9:00‑12:00 Beijing time), jointly organized by the MLNLP community, the Youth Working Committee and the Large Model & Generation Committee of the China Computer Federation.

Program committee chairs are Prof. Wang Wenjie (University of Science and Technology of China) and Associate Prof. Qian Chen (Shanghai Jiao Tong University). The event is divided into two sessions, each hosted by one of the chairs.

Invited speakers include:

Yan Yukun , Associate Researcher at Qiyuan Lab and visiting researcher at THUNLP, will present “PilotDeck: An Agent Operating System for Productivity”. The talk describes a project‑oriented architecture that binds projects to workspaces, addressing long‑task black‑box issues, context explosion, memory cost and skill drift, and enables parallel continuous execution with visual monitoring.

Ma Xianling , Professor at Beijing Institute of Technology, will present “Table Intelligence: The Foundation of Fine‑Grained Digital Understanding”. The abstract emphasizes the shift from merely processing tables to truly understanding them to unlock deeper insights.

Yu Fan , Assistant Researcher at Tsinghua University, will present “Building Multi‑Agent Learning Environments for Autonomous Learning”. The talk uses the MAIC platform as a case study to discuss generative AI‑driven adaptive education, future AI+Education scenarios, and key technical supports.

Wang Dong , Senior Algorithm Expert at Alibaba, will present “Exploring Future Directions of General Agents”. The presentation analyzes current limitations of rule‑based agents and explores six dimensions: tool/skill generation, robustness, experience reflection, dynamic memory, cost‑aware optimization, and multi‑agent collaboration.

Registration is open via a QR code and the live streams will be available on Bilibili and WeChat Video Channels.