If you’re new to dlab, first of all: welcome! This page will help you get started by providing useful information on how to get set up and ready to roll. If you’re new not only to dlab, but to EPFL as a whole, see the bottom of this page for some additional useful pointers.
As a very first step, carefully read this introduction to our dlab resources: computing, office space, etc. It will tell you everything that’s necessary to get cracking and crunching with that exciting data!
Our internal group mailing list is email@example.com. You will gain access to the mailing list together with your dlab accreditation (see instructions here). The list is our primary group-wide communication channel (in addition, there’s Slack; see below), so don’t hesitate to send email to the list frequently, even for small things. Membership in the group associated with the mailing list is also necessary to use the lab’s computing resources.
You should also subscribe to firstname.lastname@example.org, where machine-learning–related messages (mostly talk and job announcements) are being posted. To sign up, visit groupes.epfl.ch and click ‘Register in this group’.
Join the dlab team on Slack. We use Slack for casual communication that we don’t require to be archived. For conversations that you want to be able to go back to at some later point, please use email instead.
We also use Slack for brief daily updates: every day, each dlab member (PhD students, Master’s students, postdocs, interns, prof) posts a short message in the channel
#dweets (for “dlab tweets”), summarizing in a sentence or two what they’ve been up to that day. The purpose of these updates is to increase information flow inside the lab and your awareness of your own achievements (not surveillance!).
Here is some good advice on how to make meetings with your advisor as effective and successful as possible. Please read it.
Offline meetings have the great advantage of a shared black- or whiteboard to brainstorm on. Online meetings don’t, and this can make it much harder to communicate effectively. We therefore use Google Jamboard as a virtual shared whiteboard. Every lab member should create a jamboard and share the link at the beginning of the meeting. All participants can then see, and draw on, the same shared jamboard, using their finger (phone), pen (tablet), or mouse (desktop). (Install the app!) For recurring meetings, you may create a single, reusable jamboard, which can be extended indefinitely by adding new frames.
Documenting your research
It is important that you keep thorough notes of your research. This will make it easier to remember the things you’ve done, the things you still need to do, important references you’ve come across, etc. It will also let you share your thoughts with the rest of the group, and it will make you a better, more disciplined researcher.
Your notes (ideally as a Google doc), as well as any other documents you would like to share with your mentors (including Bob), should be placed inside a Google Drive folder that is named after you and shared with your mentors. Inside that folder, also create a special document called “Master doc: Your Name”. This master doc serves two main purposes: (1) It should be an index of all your other documents, which should be linked from the master doc. (2) It should contain notes for all meetings you have regarding your project (be they with Bob or with any other mentor). If you don’t have a separate document or tool for managing related work that you come across, you may also dump links to related work in the master doc.
Your shared Google Drive folder should also contain your meeting jamboards (see above).
You can find some tips and tools for writing papers in LaTeX here.
Blog posts are a great way of publicizing your research results beyond good old (and slooow…) papers. They are also a channel for making what you’re doing accessible to a general public, rather than just to other academics.
Therefore, every paper published at dlab is expected to be accompanied by a blog post that is written in non-technical terms and easily accessible to non-experts, to be published on the dlab blog (a.k.a. the dlog).
The dlab canon
Here is a list of books that every dlab member should read sooner or later.
The canon is not exclusive; in other words, you should read plenty of other books, too! Rather, it is an attempt to cover some concepts that, although they are central to our work, are unfortunately not taught as extensively in courses as some of the other central concepts we rely on (machine learning, algorithms, natural language processing, etc.).
Reading these books, in addition to paying attention in your classes on machine learning and related areas, will ensure that everyone in the lab speaks the same language. Additionally, all books listed here are a lot of fun to read; you’ll learn a lot as well as be entertained.
- Paul Rosenbaum: Observation and Experiment: An Introduction to Causal Inference, Harvard University Press, 2017.
- Judea Pearl: The Book of Why: The New Science of Cause and Effect, Basic Books, 2018.
- Matthew J. Salganik: Bit by Bit: Social Research in the Digital Age, Princeton University Press, 2017. [free online edition]
I also highly recommend chapter 2 (“Research Questions”) from Nick Huntington-Klein’s book The Effect: An Introduction to Research Design and Causality. Defining a crisp research question is often the hardest part of a project, and this chapter contains valuable reflections on how to go about finding and sharpening your research questions.
New to EPFL?
The following pointers might help you get started if you’ve just arrived at EPFL.
On campus, the best wireless network to use is the one called epfl. You can use it with your GASPAR username and password.
To access EPFL resources, such as the compute server, from outside of the EPFL network, you need to set up VPN.
During your first days at EPFL, you’ll get your CAMIPRO card, which will serve as an ID, keycard, and payment method, among other things.
EPFL Campus app
Consider installing the EPFL Campus app on your smartphone. It has useful information such as a map, people directory, restaurant menus, etc.