Here is some advice for students in my research group. I hope to add
to this over time. --Vince
Research advice
Research should be your top priority. This is what you will be evaluated
by in graduate school, and this is what will determine how successful you
are in your career as a scientist.
You should be on top of your own research and, as much as possible, be
driving it forward yourself. Research is much more fun when you feel
ownership of it. This does not mean that I will not be involved in your
research, or that you should not listen to my suggestions. But you should
not feel like I am just using you to get my projects done. You
should also be using me to get advice on the research that you
want to get done!
A not-so-good scenario:
You have a list of things that you still need to do on your research.
Every week, I make some new suggestions, adding to your list. You do not
feel ownership, and research becomes somewhat of a drag. Maybe you even
feel like some of my suggestions are not the right way to do things, but
you are not really sure what would be a better way. All of this of course
is not helping you catch up.
A good scenario:
You are pursuing your own ideas on your research. I make some suggestions;
some of them you have already thought about, and perhaps you have already
thought of something better. The few suggestions that are still useful are
now relatively easy to do, and you never feel like you are falling behind
on them. You feel in control of your research, and it is fun!
These scenarios are obviously very black-and-white, and it is not such a
terrible thing if you are in something like the first scenario for a while,
especially in the beginning of your graduate studies. I am here to help
you out when you get stuck or need ideas. But over time, you should be
trying hard to get to the second scenario. The first step to getting there
is to work hard to make sure that you do not fall behind, and that you
clearly understand your research and its context. You should be organized,
keeping track of all the ideas and results on your project, and come into
weekly meetings prepared, aware of what is going on in your project and
what we need to talk about. If you do this, you will at some point find
yourself thinking creatively about what you are doing and having fun with
it.
Doing research vs. doing other things
In graduate school, you have a lot of freedom in how you schedule
your time. Unlike most other jobs in the world, nobody is checking that
you come in at 9am and stay until 5pm. This freedom comes with the
responsibility to use it wisely. You do have a job to do, and you should
be disciplined and put in the appropriate number of hours. Try to "save"
your free time for things that are seriously fun.
Research progress can be erratic in the short run -- you may spend a week
working very hard on a problem, get nowhere, and regret that you did not
spend the week doing something else. However, in the long run, your hard
work will pay off. You did not waste that week.
Also, research progress can be hard to measure. (How important is that new
result that you came up with, really?) Working on, for example, a homework
assignment, gives a much clearer sense of progress. (You solved that
20-point question!) It is tempting to prioritize things that give a clear
sense of progress, but this can be detrimental to your research. Over
time, you will get explicit feedback about how well you are doing in
research (from the faculty, but also, more importantly, from getting your
papers published).
Some very good conferences:
If you want to know about a conference or other venue not listed here, ask
me. I may very well have forgotten some great venues. If you think I'm
wrong about something, please let me know! Some of the below is based on
very little information and these are just my limited impressions rather
than anything based on thorough analysis.
ACM Conference on Electronic Commerce (EC, also referred to as ACM-EC):
very good conference with a strong focus on auctions and mechanism design,
also some game theory and voting. Tends to have many theory papers, some
more AI-ish papers. The conference is interested in expanding to other
areas such as systems.
International Joint Conference on AI (IJCAI) and National Conference on AI
(AAAI): these are the top general AI conferences. They are generally very
good (although they tend to accept a few papers that are not so good as
well). IJCAI is perhaps slightly better overall. AAAI is the North
American conference (though researchers from all over the world publish in
it). IJCAI tends to be a little more interested in logic and complexity
than AAAI (presumably due to the European contingent), but the difference
is small.
International Joint Conference on Autonomous Agents and Multi Agent Systems
(AAMAS): an AI-ish conference but with a restricted focus, mostly on
multiagent systems. Good overall, tends to accept some weaker papers as
well, but the papers in game theory and mechanism design are generally
good.
Conference on Theoretical Aspects of Rationality and Knowledge (TARK):
conference on topics in the intersection of computer science, economics,
game theory, philosophy, and logic. Somewhat like EC but a little less
applied, smaller, and more focused on game theory and logic and less on
auctions. (Has perhaps been eclipsed by EC somewhat.) But tends to have
good papers.
Workshop on Internet and Network Economics (WINE): more like a conference
in spite of the name. Similar to EC, though EC is better. WINE is a
little bit more dominated by computer science theoreticians.
Conference on Uncertainty in AI (UAI): focuses mostly on topics such as
Bayes Nets, but usually has a couple of papers on game theory, mechanism
design, auctions, etc. as well.
International Conference on Machine Learning (ICML): focuses on machine
learning (mostly from an AI perspective), often has a couple of papers on
topics such as learning in games or learning in markets.
Neural Information Processing Systems (NIPS): mostly a machine learning
conference these days, overlap with ICML and UAI, and similarly it accepts
some learning in games, etc., papers.
International World Wide Web conference (WWW): covers many topics, but one
prominent topic is "monetization" which includes various economics-related
topics such as sponsored search, and there is also interest in social
networks.
There are also some very good conferences in computer science theory that
often publish papers on game theory, mechanism design, auctions, etc. These
include FOCS, STOC, and SODA. (FOCS and STOC are the top two theory
conferences, SODA is close.) COLT is also good for (theoretical) machine
learning.
Even more conferences, workshops, and other things:
Symposium on Algorithmic Game Theory (SAGT): similar to EC and WINE, maybe
more theoretical, less auctions and more game theory. But too new for me
to say much about it.
International Conference on Electronic Commerce (ICEC): more of an applied
focus, less game theory and mechanism design.
Innovative Applications of AI (IAAI): usually co-located (even
co-published) with AAAI, IAAI focuses on (about-to-be) deployed
applications that have AI in them.
Conference on Logic and the Foundations of Game and Decision Theory (LOFT):
similar to TARK, perhaps even more logic-heavy, no published
proceedings.
European conferences/workshops: ECAI (like AAAI), ECML (like ICML), EUMAS
(like AAMAS).
International Symposium on AI and Math (ISAIM): AI with perhaps a more
theoretical bend, not as good as the top AI conferences.
KR: reasonably good AI conference on knowledge representation.
KDD: knowledge discovery and data mining. Supposed to be good but I don't
know much about it.
Stony Brook game theory conference: yearly conference on game theory,
mostly economists, abstracts only.
GAMES: large conference on game theory held once every 4 years. No
proceedings.
INFORMS Annual Meeting: huge conference focusing on operations research,
management science, etc. Abstracts only.
COMSOC (Computational Social Choice) workshop: a nice, relatively new workshop
mostly on voting (but also mechanism design, etc.).
AMEC (Agent-Mediated Electronic Commerce) workshop: a nice workshop with
topics similar to the AI-ish topics in ACM-EC.
Multidisciplinary Preferences workshop: nice workshop that focuses on
preference representation, elicitation, social choice and mechanism design,
etc., often from an AI perspective.
Incentive-Based Computing (IBC): another workshop on game theory in cs and
the like.
NetEcon: this workshop focuses on economic phenomena in networks.
Game Theory and Decision Theory (GTDT): a nice multiagent systems workshop on
game theory
Dagstuhl workshops: many computer science workshops are organized at
Schloss Dagstuhl, occasionally on a topic relevant to us. Usually by
invitation.
Interesting competitions: the Trading Agent Competitions (TAC), AAAI computer
poker competition.
Some very good journals:
Journal of the ACM (JACM): journal devoted to the best papers across
computer science. Perhaps focuses a little more on theory papers but also
accepts AI papers.
Artificial Intelligence (AIJ), Journal of Artificial Intelligence Research (JAIR):
top AI journals.
Journal on Autonomous Agents and Multi Agent Systems (JAAMAS): the AAMAS
community's journal, a good journal.
Machine Learning, Journal of Machine Learning Research: top machine
learning journals.
Econometrica: perhaps the top economics journal; theoretical. Does not
like to publish papers that have appeared in cs conferences.
American Economic Review (AER): also perhaps the top economics journal;
less theoretical.
Journal of Economic Theory (JET): a very good theoretical economics
journal. Is open to papers that have appeared in cs conferences.
Games and Economic Behavior (GEB): the top game theory journal. Is open to papers that have appeared in cs conferences.
International Journal of Game Theory (IJGT): another good game theory journal.
(There are a lot more economics journals...)
Great operations research/management science journals: Operations Research,
Math of Operations Research, Management Science. (Again, there are
others...)
Social Choice and Welfare is perhaps the top social choice journal (though
some of the top economics journals will also accept social choice
papers).