▌ ▐ ▐▀▌ █ █▀▌ ▐▀▌ ▌ ▐ ▐▌▐▌ ▐▀▀ ▀▌▐▌▐▀ █▀█ █▀█ █ █ █ █▀█ █ █ ▐▐▌█ █▀ ▌▐▌▐ ▌ ▐ ▌ ▐ ▐██ ███ ▌ ▐ █▄█ ▐ █ ▐██ ▄▌▐▌▐▄ (W)elcome (A)dvisees Tal(k)s (P)ublications (V)ita/Bio/Contact (F)AQ (T)eaching (B)log (L)inks
Frequently asked questions:
  • Would you like to give a talk/tutorial at my workshop/conference/department/lab?

    Inclusion Rider: In general, I am honored to be invited to give talks, and especially enjoy meeting new people and discussing research. I am, however, unwilling to do so in a way that contributes to the further marginalization of womxn and gender minorities, people of color of all genders, people with disabilities, and other amazing members of our disciplines whose voices are often drowned out.

    For instance, I decline invitations for: events with all male organizers; workshops/panels with non-diverse speakers/panelists; talk series whose past speakers are non-diverse. But I will also try to assist. If you invite me and I decline, I will provide a list of alternate invitees you might consider (adapted for the topic of interest); or you can just email me for suggestions. Also check out the WiML Directory, the Black in AI workshop, and the Widening NLP Spotlights.

    This won't necessarily solve everything, and making process here is hard: I think it's important to keep in mind that service overload is a real thing for members of underrepresented groups, and we all need to find a way to balance priorities and what they ask for.

    I'd like us, as a community, to get to the point where having such "rules" is unnecessary, in which all researchers are valued equally for their contributions, and hopefully we can get there. I am partly taking a firm stance here because I am guilty in the past of not paying enough attention to these issues, and have organized my share of, e.g., all-male panels; I believe we can make the most progress when we all hold each other accountable. I'm writing this in part to hold myself accountable. If you're trying to find other ways to help, please consider the list of suggestions from Kathryn McKinley on ACM SIGARCH.
  • What's with your name?

    My family name, "Daumé" has an acute accent which changes the height of the vowel, from [ə] ("eh") to [e] ("ay" as in "may"); you can say "dow-may". The "III" (roman numeral three) is because my name matches my father's and also my (paternal) grandfather's. It's pronounced "three" like the number.

    You might be wondering how to generate my name in various software packages. It's easy:
        LaTeX:  Hal Daum\'e III
        BiBTeX: Daum\'e, III, Hal
        HTML:   Hal Daumé III
    I appreciate citations, even more so if you include my full name!
  • Will you be on my proposal/dissertation committee?

    Good question! See this document!
  • How many students do you expect to take next year?

    I like to have about 4-5 graduate students to work with at any given time. Take a look at how many students I have currently and you can figure it out. Exceptions are of course made for amazing students like you! On average, I expected to take one new student per year.
  • I'm a student at UMD; how do I join your group?

    Take one of my classes and impress me. Otherwise impress me with a software/research project you've done independently.
  • I'm a student applying UMD; how do I join your group?

    If you want to work with me, be sure you list me as the first or second choice professor to work with and list NLP as your area of choice. If in doubt, email me after you've completed your application, I can check to ensure you show up in the system.
  • How does student funding work?

    The current funding model works like this. Most incoming Ph.D. students are funded by the department for the first year (typically under a TAship). After that, you'll need to find an advisor and get fuding through an RAship. If you're good, this has never been a problem. We're unlikely to admit you if we don't think that you'll be able to get an RAship after the first year.
  • What classes should I take?

    First, I don't care about classes. I also don't know much about which profs are good and which are...not. So don't ask me. Ask your friends. I expect all of my students to know the material covered in a graduate machine learning course, a graduate NLP course, a graduate algorithms course, an undergraduate prob/stats course and an undergraduate linear algebra course. If your interests are more on the ML side, then I expect you to know optimization. If they're more on the NLP side, then I expect you to syntax and at least one other linguistics course (eg semantics or psycholinguistics or phonology or typology). Whether you get this knowledge from classes or on your own is up to you.
  • How many classes should I take?

    I would probably recommend taking 1-2 "real" classes per semester for your first year, plus a seminars. Most of the courses listed above require a fair amount of effort, so do not take 3 of them light heartedly. There's nothing wrong with putting one off for one or two semesters in favor of having time to work on research.

credits: design and font inspired by Seth Able's LoRD, some images converted to ANSI using ManyTools, original drawing of me by anonymous.
last updated on twelve may, two thousand twenty four.