Minghao Fu
Before UCSD, I completed my M.S. in Machine Learning at MBZUAI under Kun Zhang. During that time, I was also a visiting student at CMU, working with Peter Spirtes. I worked on causal learning and identification problems, with using them to improve downstream generative models and scientific discovery. My master’s thesis can be found here.
Prior to MBZUAI, I earned my undergraduate degree from UESTC, advised by Jie Shao, and conducted research as an intern in the NLP group at Shanghai AI Lab, led by Lingkong Peng.
Research
My research centers on causal-driven world models that bridges causal representations, generative modeling, and reinforcement learning, aiming to build interactive agents that learn the world, understand the world, and act through causality rather than correlation.
Publications & Preprints
Towards Identifiability of Hierarchical Temporal Causal Representation Learning
Online Time Series Forecasting with Theoretical Guarantees
Learning Task-Sufficient World Models via Intervention-Curriculum Co-Design
Learning General Causal Structures with Hidden Dynamic Process for Climate Analysis
From Comparison to Composition: Understanding Machine Cognition of Unseen Categories
How Effective is Your Rebuttal? Identifying Causal Models from the OpenReview System
Infant Cry Detection Using Causal Temporal Representation
Education
University of California, San Diego — Ph.D., HDSI
Carnegie Mellon University — Visiting Student, CLeaR
Mohamed bin Zayed University of Artificial Intelligence — M.S., Machine Learning
University of Electronic Science and Technology of China — B.S., Software Engineering
Service
Conference: ICLR 2025; NeurIPS 2025; BMVC 2023.
Journal: IEEE Transactions on Image Processing (TIP) 2024