Hello people! This page contains some answers to questions that I find myself frequently providing. I record these answers here in case they are pertinent to a question you have. If you have other questions, feel free to email me.

**Q: I am interested in applying to a computer science degree program at Columbia. Should I do so and what are my chances of getting in?**- Please see the CS department’s admissions webpage. There, you’ll find information about all aspects of the application process (e.g., deadlines, your chances of being admitted), as well as information about the academic programs themselves.
- Some frequently asked questions, already answered for your convenience, can be found by clicking the following links:
- Ph.D. applicants: Please mention my name
in your application if you are interested in my research. This is a good
way to draw my (and others’) attention to your application.
(There should be a box you can check or something like that …)
- In general, my Ph.D. students tend to (1) be interested in working theoretical aspects of algorithmic statistics and machine learning, (2) have taken some advanced undergraduate courses in math, statistics, and/or theoretical computer science, and (3) have some research experience that is discussed by a mentor/advisor in a recommendation letter.
- Note that having co-authored a paper appearing at NeurIPS/ICML/etc.
is
**not necessary nor sufficient**.

- I get a lot of emails about admissions, and unfortunately, I cannot write a personal reply to each one.

**Q: Do you have any internship positions available?**- I do not have internship positions available (virtual or in-person).

**Q: Do you have any open postdoc positions?**- I do not have any open postdoc positions. I’ll update this page if this changes.

**Q: Can I do a research visit with you for***n*months? My company/government/self will cover all of my expenses; I will not require a stipend.- Due to time constraints, I cannot commit to hosting external students or visitors whom I’ve never met or worked with.

**Q: What is your advising style? What is it like to be your Ph.D. student?**- I don’t have a fixed advising style; I’ve found that no one mode of advising works for all students. My website has a list of my current (and past) Ph.D. students. You are welcome to contact them and find out about the range of advising styles that I have employed.

**Q: I am interested in taking a course in and/or doing research in machine learning. What courses should I take in preparation?**- First, it’s great that you are interested in these subjects!
*Machine learning*is a confluence of ideas from many disciplines, including computer science, optimization, physics, and statistics. Because of this, it is a very broad subject that builds on a number of foundations. When getting started in machine learning, it may feel overwhelming—it did for me. At the same time, I hope it also means that there’ll be something particular that’ll “click” with you and that you’ll want to study in depth. - A good way to get started is to build up solid mathematical
foundations in
*linear algebra*,*probability*, and*multivariable calculus*.

- Comfort in writing code to process and analyze data (e.g., in Python) is also very helpful.
- The syllabus for COMS 4771 “Machine Learning” has some resources on these prerequisite subjects.

- First, it’s great that you are interested in these subjects!
**Q: Same as above, but particularly in learning theory…***Learning theory*is the mathematical study of models and algorithms for learning. There are several courses related to learning theory at Columbia, including COMS 4252 “Introduction to Computational Learning Theory”, COMS 4773 “Machine Learning Theory”, and IEOR 8100 “Reinforcement Learning”.- These courses are “proof-based”, in the sense that the course is largely driven by formal definitions of and mathematical theorems about learning problems and algorithms. The proofs and the intuitions behind these theorems help solidify our understanding of the phenomenon that we call learning.

**Q: What are good books to read about machine learning and/or learning theory?**- For general machine learning:
*A Course in Machine Learning*by Hal Daumé III. - For learning theory:
*An Introduction to Computational Learning Theory*by Kearns and Vazirani;*A Probabilistic Theory of Pattern Recognition*by Devroye, Györfi, and Lugosi.

- For general machine learning:

**Q: I am currently a Columbia undergraduate/MS student; will you supervise my independent research project or thesis?**- Due to time constraints, I can only commit to supervising project/thesis students who have done well in a class I’ve taught, or who have done well in a relevant class (e.g., in learning theory) and the instructor can send a strong letter of reference (an informal email is enough). Depending on your interests, I may recommend some courses that I think are useful to take before diving in.

**Q: Do you have any open research assistant positions available in your lab?**- No, I don’t even have a “lab”.

(These are not frequent answers to questions, but rather just links to useful “advice” that others have written.)

- Mihir Bellare’s views on the Ph.D. experience.
- Fan Chung’s advice for dealing with a math bully and general advice on research for graduate students.
- Jeannette Wing’s tips on the interview process (for academic jobs).
- Charles Isbell’s thoughts on getting admitted to a Ph.D. program.
- Aaditya Ramdas’s checklists for Stat-ML PhD students.
- Austin Z. Henley’s lessons from PhD.

- George Gopen’s and Judith Swan’s article, “The Science of Scientific Writing”.
- Notes from a course on mathematical writing by Donald E. Knuth, Tracy Larrabee, and Paul M. Roberts.