Questions 

Here are answers to questions I frequently encounter

• Do you take students now? No. I haven't started any official appointments at UW Madison CS. So, even if you contact me, I unfortunately cannot help so much with your admission. But, I might be able to introduce you to other faculty members depending on your research interests. The CS department is rapidly growing, and the current ranking in US news is 12th.  

• Do you belong to an ML theory community?  I have never labeled myself as a learning theory person. My interest has always been super forward-looking, i.e., my goal is to develop innovative & practically useful ML algorithms. But, I use theoretical thinking as a strong inductive bias to achieve this goal with a solid foundation. 

• Are you a statistician or a computer scientist?  My current main research interests lie in ML, AI, and its application in science. But, I still love reading papers in statistics journals. So, statistics would still be my lifetime hobby, and I believe it is an excellent tool to achieve the abovementioned goal. 

• What are good resources for reinforcement learning?  While there are so many resources in RL, the below two are sufficient to start research:

Wen Sun's course:  https://wensun.github.io/CS6789_spring_2023.html

  This is a fantastic resource to learn the fundamentals of reinforcement learning.  

Sergey Levine's course: https://www.youtube.com/channel/UC4e_-TvgALrwE1dUPvF_UTQ/videos

This is a fantastic resource for learnning deep reinforcement learning.

• What are good resources for causal inference?  

Causal inference I am referring to here is more statistical aspects (i.e., semiparametric efficiency, doubly robust estimators, offline policy learning, IV, DID, RDD, etc.). I strongly recommend this book 

Applied Causal Inference Powered by ML and AI by V. Chernozhukov, C. Hansen, N. Kallus, M. Spindler, V. Syrgkanis

 I also learned a lot from the materials from Andrea Rotnitzky and Vasilis Syrgkanis. It looks like these materials are not in public 

+ Causality roughly consists of (1) causal inference, (2) causal identification (i.e., like ID algorithm), and (3) causal discovery. I am not familiar with (3). Regarding (2), I have learned a lot from papers by Elias Barenboim and Ilya Shpitser. But, again, I am not sure about a very instructive material that is good for beginners. 

• What are good resources for learning the foundations of statistical machine learning?  

    Of course, you don't need to know all of the stuff below to start research. Definitely, attempting to learn everything is overkill (at least, I still don't fully grasp many parts yet). But, it would be fun and meaningful to learn gradually. 

    (Books) 

High-Dimensional Statistics by Martin J. Wainwright: Bible of statistical learning theory.  

Asymptotic statistics by  A. W. van der Vaart:  Bile of mathematical statistics. Learning this might be a bit tough in the era of deep learning. There are many distractions in the world now:) But, I have learned a lot from it.

Bandit algorithms by Tor Lattimore and Csaba Szepesvári: Bible of bandit/online learning

Introduction to Online Convex Optimization by Elad Hazan: This is short; but extremely insightful and organized material for online learning

    (Cousres online)

     A course offered by Ryan Tibshirani

     High-Dimensional Statistics