Mentor: Da Yang, Assistant Professor, Geophysical Sciences
William Zheng is a third-year undergraduate majoring in Physics and Statistics. He has a strong interest in machine learning and reinforcement learning, particularly in their applications to physics research.
As a Polsky Research Fellow, Zheng used Machine Learning methods to investigate the major source of predictability of Madden-Julian Oscillation(MJO), which is an eastward-moving tropical atmospheric disturbance that have significant impact on precipitation. Zheng distilled existing neural networks to achieve a simple model with 3 week predicability using less than 100K parameters and will continue to work on improving the interoperability of the model.
“This internship strengthened my foundation for a future career in climate by giving me hands-on experience with the Madden-Julian Oscillation (MJO) forecast. I developed skills in analyzing large-scale OLR datasets, building machine learning models, and interpreting model mechanism. These tools and insights will be invaluable as I continue pursuing research in climate science.”