Research

Why do research?

Advance the frontier of knowledge.

Researchers get a sneak peak at the future. Computer science, being a relatively young field, is full of opportunities for exploration. Young researchers are capable of making a large impact on the frontier of computing thanks to the sheer extent of what remains unknown in CS.

Preparation for research-based graduate programs.

PhD and research-based masters programs focus on producing original research. By conducting research in your UG years, you demonstrate your ability to conduct research in graduate school.

A stepping stone for industry.

Research will likely expose you to some industry-relevant technical knowledge. Examples range from practical experience with a specific tool (e.g.: a programming language, a machine learning library) to in-depth knowledge of an industrially relevant domain (e.g.: software security, natural language processing).

I claim that research is a strong alternative to personal projects. Although supervised research doesn't provide the same freedom to choose what you want to create and what technology stack you'll use, you will get more mentorship and work in a team.

Earn credits or $$$!!!

You can take CS1950 to fulfill your capstone, write and defend an honors thesis to earn a Bachelors of Philosophy, or get paid in cold, hard cash as a research assistant or intern!

Getting started.

Getting your foot in the door might be intimidating: the ivory tower isn't always an accessible place. Luckily for you, there is practically always more demand than supply of research labor. This section describes several possible pathways to get into research.

Pitt CS undergraduate research symposium.

Pitt's CS department hosts an annual research symposium in the fall. Professors give short presentations to pitch their research projects and recruit undergraduate researchers.

You will have a chance to ask questions in a Q&A session. If you're interested in a professor's pitch, I suggest asking a question or two; asking questions demonstrate interest and makes you more memorable, which might be helpful if you later ask to join the professor's research project.

Doing well in a professor's class.

If you perform well and/or demonstrate research potential in an advanced class, the professor might invite you to conduct research with them. Demonstrating research potential is easier to achieve in classes that feature a research-y project as part of the curriculum (e.g.: Human Language Technologies, Deep Learning, some graduate classes).

CS1980 (not to be confused with CS1950)

Sometimes research projects show up as one of the project options for the CS1980 capstone course.

Cold-emailing professors.

Suppose that you're considering doing research with Professor X.

First decide whether Professor X's research interests and/or current projects are a good fit for you. If you had taken a class or attended a talk by Professor X, then you probably already have an idea of their interests, although I would still suggest digging a little deeper.

Go to Professor X's website and look for a description of their research interests; this information might be on their homepage, in their CV, or on a page dedicated for research.

To get more specific information on the professor's recent research interests, I would suggest skimming over their recent publications (up to 3 years is my suggestion). Look at the titles and maybe the abstracts of their recent publications; reading entire papers is probably overboard for now (unless if you're hooked, then read on!).

If you believe that Professor X is a good fit, then send an email to signal your interest. Mention why you are interested in Professor X (e.g.: interest in their research direction, their recent projects, you liked an advanced class that they taught, you liked a lecture that they gave, et cetera), and if applicable, mention any qualifications that make you well-suited for working on a particular project or area of research. Offer to set up a meeting and/or to drop in during office hours to discuss potential research projects.

REUs (for US citizens and permanent residents)

Many universities host REUs (research experience for undergraduates). An REU is similar to an internship, except that you're working on research at a university. REU positions are salaried (usually $4000-$6000 for a summer), often come with university housing, and often invite you to seminars and talks that can help with professional development as a scientist.

Here's a search tool for finding REU programs.

REU programs generally have an overarching high-level focus (e.g: IoT, computer systems, software engineering, algorithms) and a list research projects or areas that are specific to a professor.

When applying to an REU, convince the decision-makers of the following:

  • You are interested in 1+ of the REU's research areas.
  • You are technically qualified to conduct research in 1+ of the REU's research areas.

Applications require up to 2 letters of recommendation from faculty members. If you already have research experience with a professor, then definitely ask your research supervisor to write a letter of recommendation. REU programs, however, generally understand that many applicants will not have research experience (students might come from a small school that's focused on teaching, or their school might not conduct research in their area of interest).

Other non-Pitt research programs.

Various foreign universities host research internships that accept international applicants. Examples include:

Some other programs:

Do it alone.

There is no obligation that you must do scholarly research under the guidance of someone more senior. A research advisor, however, is a tremendous resource in conducting research, and I would counsel against taking on a large research project without mentorship. The path to knowledge lies somewhere within an infinitely large search space, and a good advisor will help you traverse through the space and prune out false steps. In other words, research is hard, and mentors make research easier.

Reading research papers.

In the course of your research, you may need to read research papers. Instead of giving advice myself, I invite you to read these articles:

Remember that the above articles provide advice, not mandatory rules. Do what works best for you.

Publicizing your research.

  • Reasons for getting the word out.
  • Solicit opinions and advice. Presenting your research to an audience is an opportunity to solicit advice for improvements and future research directions.
  • Network with the research community. The people you meet while presenting your research might become future collaborators, research advisors, company references, et cetera. Some research conferences have industry sessions where you get to network with companies and get a foot in the door for a job or internship.
  • Add to your resume. Research presentations–whether in the form of a paper, poster, talk, et cetera–serve as a evidence and recognition for your research, which can augment your resume. If you're applying to research-oriented graduate programs, then this evidence of research accomplishment further demonstrates your ability to produce presentable research.

Pitt CSC talks.

You can schedule a talk at a club meeting!

Pitt's research exhibitions.

Pitt hosts various research exhibitions and poster sessions at various times of the year.

Honors college thesis.

Pitt Honors College offers the BPhil degree, which requires submitting and defending a thesis.

Conference and journal papers.

Conferences are the primary venue for publishing scholarly CS research (this is different from other scientific disciplines, where journals are typically the primary venue). Conferences and journals employ peer review, where reviewers provide feedback and decide whether to accept a paper for publication.

Terminology sidenote: in conferences, reviewers are members of a program committee.

If a conference paper is accepted, one of the authors is expected to give an oral (PowerPoint) presentation and engage with the audience in a Q&A session. There are exceptions to this rule: at some very large conferences (e.g: NeurIPS, ICCV), an author presents a poster instead of an oral presentation.

Beware of predatory conferences and journals. They have low publishing standards and charge ridiculous fees (that's why they exist). If you get an unsolicited email asking for you to submit to a conference/journal, there's a decent chance that the venue is predatory.

Conference posters and short papers.

Some conferences also accept posters or short-paper publications. Posters and short papers generally have higher acceptance rates than full-on research papers. Moreover, it's generally more acceptable to submit in-progress research and early results in posters and short papers.

Typically, if you want to present a poster, then you need to actually first write and submit a short paper or extended abstract, rather than submitting a digital version of a poster. The conference will decide whether to invite you to present a poster based on the contents of your written deliverable.

Workshops.

Academic workshops are similar to conferences: they feature posters and/or oral presentations. Workshops differ from conferences in term of size–they're smaller–and scope–they tend to focus on a more narrow topic.

Workshops, like conference posters or short papers, are also more welcoming of in-progress research and preliminary results, and their acceptance rates are also often higher than full conference research papers.

The narrow topic of workshops often attracts that topic's forerunners, giving the workshop attendee a very good opportunity to network with important people.

There are exceptions: some workshops in theoretical CS are as competitive as top conferences.

Student research competitions.

Some conferences host student research competitions. Generally, the process is as follows:

  1. Submit an extended abstract or other short written deliverable.
  2. If the program committee accepts your abstract, then present a poster at the conference. Judges will talk to you about your poster and ask questions.
  3. If judges like your poster, then give an oral presentation and have a Q&A with judges.
  4. If judges like your presentation, win glory, prizes, and $$$ in a big awards ceremony!

Undergraduate and graduate students compete separately, so don't worry about getting pwned by PhDs with 5 years of experience.

arXiv.

arXiv is an online repository of preprints–early, non-reviewed scholarly papers. Online preprints facilitate rapid sharing of new research. Some fields (e.g: machine learning) upload dozens of preprints per day onto arXiv.

In addition to sharing preprints, you may also use arXiv to upload complete research that is difficult to publish (e.g.: negative results, which are harder to publish in conferences or journals).

If you intend to submit an article to both a conference/journal and arXiv, then check the conference's/journal's policy on uploading to preprint servers. Some prohibit concurrent submissions to both.

Last updated: Jul 15th 2020