I do computational design for Digital Twins that learn, decide, and adapt in real time.
I develop AI-enabled methods for real-time decision-making and design optimization under uncertainty — integrating physics-based models, AI, and design methods to engineering and optimize intelligent systems.
A dad, a husband, a Christian, and a researcher in design optimization and Digital Twins.
Originally from Kaohsiung, Taiwan, I'm a postdoc in Mechanical Engineering at Northwestern University, advised by Dr. Wei Chen in the Integrated DEsign Automation Laboratory (IDEAL). My work integrates physics-based knowledge, machine learning surrogates and uncertainty quantification to enable efficient, reliable, and online-adaptable decision-making for Digital Twin systems. I will join the Department of Mechanical Engineering at National Taiwan University as a tenure-track assistant professor starting from February 1st, 2027.
My research applies data-driven methods to the core challenges of Digital Twins and design optimization under uncertainty — integrating machine learning with physics-based models as a surrogate for complex physics, and developing algorithms for real-time, online-adaptable, intelligent decision-making. It spans four connected streams.
Digital Twins are more than virtual replicas. They enable bi-directional interaction between a physical system and its digital counterpart. Integrating AI, uncertainty quantification, and real-time data, This thrust aims todevelop methods for system design, performance optimization, and lifecycle management with the aim of improving decision-making under uncertainty, including model predictive control, reinforcement learning, and LLMS.
Read more →
Sequential decision-making is the key to active data collection for many scientific breakthroughs, such as material discovery and autonomous lab. The key technique is how to leverage quantified model uncertainty to guide the selection of the most informative samples. This thrust aims to design active learning algorithms that maximize sampling efficiency for cost-expensive simulations, experiments, or multi-fidelty datasources for surrogate modeling, high-throughput experimental design, and system optimization, supporting autonomous design.
Read more →
We develop data-driven surrogate models for complex engineering systems where high-fidelity simulations are accurate but too expensive for design exploration, optimization, and real-time decision-making. This thrust advances methods such as Gaussian processes, operator learning, and graph-based learning to emulate spatio-temporal physics, quantify uncertainty, improve interpretability, and bridge simulation-to-real gaps.
Read more →Yi-Ping Chen in bold. Click any paper for the abstract, figure, and links. Full list on Google Scholar.
Conference talks & presentations are listed in the Talks section below.
Peer-reviewed full papers presented at conferences. Click any paper for details. Yi-Ping Chen in bold.
My full CV is embedded below. You can also download it or open it in a new tab.
I'm on the academic job market and always glad to talk research, collaboration, or mentoring. Reach out any time.