Kangning Diao

Hello! My name is Kangning Diao (刁康宁), and I am currently a postdoc at BCCP, UC Berkeley working with Prof. Uros Seljak. I obtained my Ph.D. under the supervision of Prof. Yi Mao at Tsinghua University, with a focus on statistical methods in 21 cm cosmology. My research spanning from 21 cm signal, Galactic synchrotron to weak lensing and exoplanets, with a particular emphasis on employing statistical methods. In terms of programming, I am proficient in Python and frequently utilize packages such as JAX and PyTorch.

I am a cautious person, and I believe my personality drives my research taste. I am particularly interested in statistical validity. Under this broad scope, my work spans from classical methods to AI applications. For the former, I am working on the statistical validation of exoplanet transit light curves using persistent sampling. For the latter, I developed a Microcanonical Langevin Monte Carlo-based training method, SMILE, to train Bayesian neural networks that are aware of their own uncertainties. Additionally, I recast the problem of finding new effects in observations as an out-of-distribution detection problem, and am working on detecting those with generative models such as Flow Matching. Preliminary results can be found in this paper.

Outside of my academic pursuits, I have cultivated several hobbies. As a Tsinghua student, I prioritize physical fitness and actively participate in Taekwondo as a member of the university team. I dedicate an amount of time each week to exercising and maintaining a healthy lifestyle. Additionally, I have a keen interest in playing the guitar, although I still consider myself a novice. Lastly, I thoroughly enjoy cooking, particularly Chinese cuisines. If you ever have an interest in sampling some of my culinary creations, feel free to reach out to me!

On the internet, I have a keen interest in aviation and electronic games. My favorite plane is the Boeing 787, and my favorite airport is PKX. I also admire the COMAC C919 and hope it achieves great success. As a fan of JRPGs, I’m currently playing “Honkai: Star Rail” by Mihoyo, although my all-time favorite game is “Persona 5 Royal”.