Staying Current
Even if it seems to take a lot of time, it is very important to know
what the other people are doing in your area. Below are some suggestions
on how to do it efficiently.
Email lists, Newsletters, Discussion Forums
There are a few specialized email lists, newsletters, and discussion
forums that you may want to follow. People sometimes announce papers
there, and sometimes there are interesting discussions on current
topic--they also allow you to get to know how some of the people think
in the area.
One prominent one is the connectionists email list for established
researchers in neural networks. In evolutionary computation, there's
the
EC-Digest
and SIGEvolution
newsletters. IEEE Computational Intelligence Society, Cognitive
Science Society, and International Society for Artifical Life each
have their own as well. Googlegroups on cigames, ml-news, and rl-list
may be relevant, although much of the traffic consists of job
announcements.
Finally, there are Reddit groups on machine learning and data science,
and facebook group on deeplearning---and there's hackernews and kaggle
forums. They have a lot of noise though.
Literature
You should follow the current contents of the most important journals
and conferences in your area. The contents are usually posted on the
appropriate email lists and newsgroups, and you can always check them
through the journal and conference web pages as well. There's
certainly no point reading every article that comes out, but when you
see an article that is especially interesting, you can usually find a
copy of it on the web. Even if they are behind a paywall, UT Library
has subscriptions to most of them, and you can get them through the
library website or through ezproxy (see the
literature survey for instructions).
There are also various services that send you customized alerts on
journal and conference contents. I don't have much experience with
them (and they change all the time), but you may want to look into
setting something like that up.
Conferences
If you follow the literature and the email lists, you'll have a pretty
good idea what is going on. But you still have to go to
conferences. Nothing compares to talking to people about what they are
planning to do next, where they think the field is going, and what they
think is really exciting right now and what is not. You have to
talk to neurobiologists about the plausibility of your model, and
speculate with them about the possible causes of behavior; you cannot
get the same perspective from journal articles.
Usually if your research is progressing well, you can put together one
or two conference papers each year, and that way get to attend
conferences as a presenter. However, even if you do not have a paper
to present, major conferences like those listed in
GoingToConferences, or
conferences that happen to be in the same geographic area, may be well
worth it. If there happens to be a specialty workshop in your area in
these conferences you should consider going (even though workshop
papers aren't necessarily prestigious publications), because work
presented there is usually very current, and people are generally very
accessible in such workshops. See the
GoingToConferences page for the
most important conferences and general tips on how to do it right.
Networking
Going to conferences is a the most obvious way of getting to know the
people in your field, and making sure they know you. However, it is a
good idea to make a more serious and systematic effort to network with
them. The idea is not so much "trying to impress the right people", but
more to recognize that science is done by people, and your work is a lot
more fun and more productive if you maintain an ongoing discussion with
your colleagues. It may involve emailing them about your recent paper,
or comments about their recent paper, or stopping by and perhaps
giving a talk when you are in town.
Last modified: Sun Sep 25 15:38:20 PDT 2022