Dear Prospective MS and PhD Students,
<BR><BR>
I get many requests a day from students all over the world wishing to
be admitted into our CS program (Undergrad, Master's and PhD). I also  get general questions
about my research lab and our projects. You can find out
all you need to know about our research from the pages and papers linked off of my home page and www.cs.columbia.edu/learning. It is unlikely that I can give you more general information about our
research via telephone or in person visits because that would take too much time to do fairly for all those interested. If you have specific questions about a paper, please email my coauthors first or me if they do not respond. Please also note that I typically only have funding to admit 0-1 new students
each year and the admissions process is quite competitive. If you
email me about your application directly, I might not be able to
respond (or may take a long time) because I get too many such emails a
day. I apologize for this but it is a reality.<BR><BR>
That said, I rank
applicants by their prior publications in conferences and journals I
care about (NIPS, ICML, UAI, AISTAT, CVPR, ICCV, PAMI, JMLR and
the Machine Learning Journal). If you don't have any such publications,
you can still apply but those who have published will
obviously appear much stronger in our rankings. I also look for students <B> <I> with extremely strong mathematical ability</I></B>. Other factors we take into
consideration incude where a student has obtained prior
degrees and with whom they have worked with in the past. If you have a
strong track record in research, I will do my best to contact you and
get you a position however you still have to apply to the CS programs
as usual and indicate interest in working with me and my research
group.
<BR><BR>
If you are admitted to do a PhD with me in my group, I will ensure your funding (for a reasonable number of years). However, most students in the MS program do not obtain funding. I work very hard writing proposals and reports to government to obtain funding to support graduate students in my lab. However we
rarely get enough funding for everyone and it is always accompanied by
a large amount of variability and uncertainty since funding decisions
and amounts are up
to government agencies and not under my immediate control. In addition,
funding is given in exchange for applied work and deliverables that we do for the agency. If there
is funding available, I usually have to distribute it among many
worthy MS/PhD candidates in my group and sometimes there is not enough
for everyone. Under those circumstances, funding priority goes to
those who produce the most work, in terms of publications 
and deliverables for agencies that support us.
<BR><BR>
If you believe there should be more funding for computer science
graduate students, feel free to write to your representatives in
congress and explain these circumstances. Funding for university
research and computer science has seen a relative decrease in the past few years. This is
unfortunate since almost all of these funding dollars go directly to brilliant graduate students and researchers for their
tuition and stipend to support their hard  work as they produce great research (like the internet itself)
that eventually gives so much more back to society than most other investments.
<BR><BR>
Good luck!
<BR><BR>
Tony Jebara
<BR><BR>

p.s. If you want to work with me without funding (either for credit or
just for general interest), you must first take 4771 and do
well. Otherwise, it will be too difficult to get a project going.
Also, you cannot TA a course with me unless you have already taken it
and done well (or are a PhD student in my group).

<BR><BR>

p.s. If you need a letter of recommendation from me for graduate
school applications, you must first take 4771 and 6772 and obtain A or
A+ in both. Do not ask me for a letter more than 2 years after having
taken a course with me since I will not remember your projects and
your work well enough to write. I'm always happy to write a letter if
you have worked on and published a paper with me. Otherwise, no
letter.

<BR><BR> 

p.s. If you need a letter of recommendation for job applications or
fellowships, we must have worked on and published a paper together.
Otherwise, no letter.

<BR><BR>

p.s. We don't take on summer internship students. It is impossible to
get a publishable project going in my field in one summer unless you
already have papers in the top ML venues and are a fairly advanced PhD
student. If you email me regarding such a position, I will most likely
not reply.