Astrophysics, Machine Learning and More
I'm a postdoctoral researcher at the Berkeley Center for Cosmological Physics at UC Berkeley with expertise in machine learning, simulation-based data analysis and Bayesian inference. In my research I develop and apply data analysis techniques that can extract relevant physics from the large, high-dimensional and non-Gaussian data provided by modern astrophysical experiments. One example is weak gravitational lensing. Upcoming measurements of this effect promise to reveal new insights into the fundamental physics of our Universe. To extract this information it will be crucial to develop new analysis methods that can optimally characterize the small scales resolved by high-precision experiments.
Apart from my work in astrophysics, I am also broadly interested in solving challenging data analysis problems in the physical sciences and beyond. In spring of 2020, collaborators and I used machine learning to provide one of the earliest and at the time most accurate estimates of the COVID-19 fatality. I am further passionate about topics such as climate science, geology and urban and transportation planning.
In recent research I have developed a differentiable simulation code that produces accurate, fully non-linear weak lensing maps at low computational cost. Codes like this pave the way to a new era of weak gravitational lensing data analysis. In earlier research, I have explored corrections to lensing observables due to inaccurate theoretical modeling. As a result, I have identified an important bias to CMB lensing measurements that future and current experiments will have to correct for.
During my postdoc at UC I have specialized in machine learning and created ML-based algorithms for scientific applications. The methods I develop exploit the powers of machine learning and deep neural networks while meeting the needs of the scientific community: reliable uncertainty estimation, little fine-tuning and fast and easy training. To achieve this, I combine the best of two worlds: traditional (Bayesian) statistics and deep learning. My latest algorithms can be used for accurate anomaly detection, posterior analysis in high dimensions and artificial data generation. Browse my recent publications to learn more.
MADLens, fully differentiable lensing simulations
MADLens is a python package that does not only produce accurate non-Gaussian lensing maps at low computational cost, it is also fully differentiable with respect to cosmological parameters and the initial density field.
Probabilistic Auto-Encoder (PAE)
The PAE is an easy-to-train deep generative model that produces state-of-the art results in sample quality and outlier detection accuracy.
Deep Uncertainty Quantification (Deep UQ)
Deep UQ makes high dimensional posterior analysis tractable by combining traditional posterior analysis with machine learning.
Estimating the true fatality rate of COVID-19
A sophisticated time series analysis of Italian mortality data allowed us to accurately infer the fatality rate from COVID-19 shortly after the pandemic had begun. Since then many other publications have confirmed our findings.
I am committed to making all of my codes user-friendly, well documented and publicly available under creative common licenses. All of my codes are hosted on GitHub. Feel free to contact me for support or to report bugs.
I offer research projects to students of all levels and I typically mentor 2-4 students at a time. Because of the upcoming application season I am currently (status fall 2021) not accepting new students, but stay tuned for updates!
Students who are interested in research projects at the intersection of ML and (astro-)physics are welcome to contact me. Programming experience, preferably in Python, is required.
Lister is majoring in Astrophysics and Applied Maths. She is currently in her junior year.
Project: Weak lensing posterior analysis
Max E. Lee
Max is pursuing a double major in Physics/Astrophysics and currently in his senior year.
Project: Developing fully differentiable lensing simulations
Campbell Hall 341
University of California
Berkeley, CA 94720