Ms. Yaonan Gu

PhD Candidate

National University of Singapore

With the rapid development of AI technologies, their applications have revolutionized various fields such as energy, healthcare, education, entertainment, and social sciences. In the energy sector, building energy consumption accounts for over one-third of the world's total energy usage, making energy efficiency a critical priority. Precise building energy simulation models are essential for improving energy performance and reducing carbon emissions. However, these models often incorporate uncertainties from various sources.

 

My research focuses on Bayesian calibration techniques, which leverage prior domain knowledge to quantify and reduce model uncertainties, thus enhancing the accuracy of energy predictions. While Bayesian calibration offers significant advantages, existing frameworks in the energy domain are limited to unrealistic single-output models and are computationally expensive. Additionally, the influence of data streams on model performance has not been fully explored.

 

To address these challenges, my research aims to develop more efficient Bayesian calibration algorithms that consider multi-output scenarios and reduce computational costs. Furthermore, I seek to identify the impacts of data streams on model calibration performance. Building on my experience with large language model (LLM) algorithms during my research internships at Tencent and Baidu, I am also integrating these LLM techniques into my future research trajectory.