My first conferences as a PhD student

Last week I presented at two conferences. I have attended many conferences while I was at Google. I also presented my work at an internal research conference at Google. But I had never presented at any external conferences before. Last week was the first time. I was excited about these presentations. The first one was here in Montreal called “Optimization Days 2022” on 16-18 May. It was organized by GERAD (Group for Research in Decision Analysis). The second one was at Quebec City called Mobilit.ai on 17-19 May. In this post, I will walk you through the experiences.

Registration

This part was simple for optimization days. I had to submit an abstract for the talk. I was a bit concerned about whether they will accept my presentation or not. But it turns out that they accept almost any presentation that is related to operations research and optimization. So, my abstract was very quickly accepted. At that time, no sessions were created. So, my presentation was supposed to be added to some session later. My only concern was that the other conference had overlapping dates. I was told that my presentation at the Mobilit.ai will be on 18 May. So, I requested the organizers of Optimization days to schedule my presentation on 16 or 17 May.

For the mobilit.ai, I didn’t need to register. The organizers approached Andrea and since he was unable to present due to his schedule, he asked me. That’s how I ended up presenting there. Andrea thought it would be a good opportunity to show my work. I am also a part of DEEL (Dependable and Explainable Learning) Quebec, who basically organized this conference. My work had a lot to do with explainability. So, it made sense that I should present it there. I used the same abstract for this presentation as well.

Preparation

Initially, I planned to create the presentation a week before the conferences. But it turned out that I needed to submit my presentation to the industrial partners for a review. Therefore, the organizers asked me to submit the first draft by May 2. This was a good thing. While creating the presentation, I thought about what questions my audience will ask. That brought some more clarity about my work. I even thought about trying some simple tricks to improve the results a little. When I started working on those things, I found more. The results had significantly improved (by 4 folds) compared to the results recorded in the first draft. My confidence increased with the better results. Now that I think about it, this improvement was only because I was presenting. If I didn’t have to present my work, I wouldn’t have worked on these ideas.

Meanwhile, I was continuously improving my speech plan and the slides. I had rehearsed my presentation a lot. The total time for the optimization days presentation was 25 minutes. This time included the questions from the audience as well. So effectively, I had 20 minutes. I had to prepare a presentation that can be finished in 18-19 minutes to have a safety block. This made it a bit hard. I had to chop down many details. But it was a good thing. This made my presentation less technical and easy to follow. I also looked for presentation tips and used some of them to improve my slides and the structure of my presentation.

Presentation at Optimization Days

Since I submitted my abstract early, my presentation was not part of any sessions. The sessions were created after my submission. The organizers had to create a separate session to accommodate multiple presentations that were submitted early. My presentation was part of one such session called “column generation”. This is a heavy title for my work category. To add a cherry on the top, they made me the session chair. Isn’t that amazing? I am a session chair in my very first PhD conference! That too of an important session.

I wasn’t sure how many people would attend this session. There were many other sessions going on in parallel. I was hoping to not get embarrassed by a tiny audience showing up. But many people showed up. The room looked about ¾ full. My talk was at the end since I was the chair. Some people left the session before that because my other advisor Guy was presenting at the same time in a different session. Well, I can’t do anything about it.

The room was still quite full. I presented my work. Things went more or less according to the plan. Although, at the end, my friend told me that it was a bit fast. I could see that many people were unable to follow. But sill, many people could follow and asked questions at the end. I was hoping to get some suggestions as well, but that didn’t happen. The questions were straightforward.

My talk was on the first day, first session. That gave me a lot of time to relax before my next presentation at mobilit.ai. I attended some other sessions. Some of them were quite interesting and directly related to my next projects. I got to interact with some researchers doing related work. It was a good experience overall.

Presentation at Mobilit.ai

The time limit for this presentation was a bit higher. I had about 27 minutes to present my work, and that didn’t include audience questions. I could present more details. But instead, I decided to add more elaboration to my existing presentation.

This presentation was live-streamed. The organizers showed me all the tools for the presentation before my talk. The projector screen was huge. I had two large monitors in fort of me to access my notes. There was a countdown timer in front of me. I had a clicker. They put a mic that went behind my head across the ears with a controller hanging at the back on my belt. Kind of like the one TV show hosts use. I felt like a celebrity. I was introduced by the host before my presentation. This introduction was in French. But I understood most of it because it was the translation of the English introduction I submitted to the organizers.

The talk went well. This was mostly bilingual audience who spoke French and English. The conference was mostly in French, but a few talks were in English. The room was too large for me to make eye contact with everyone. I tried my best. I could sense that a few people were following. Not a lot of people here had a background with operations research. But they were still interested in my work because it had a lot to do with “explainable machine learning models”. After the talk, many people approached me and discussed ideas. They all liked my work. Some of them were doing work very similar to mine. They faced similar issues, and my presentation helped them a little. Hearing about their approach gave me some new ideas that I can use in my work as well.

My presentation here gave me a lot of attention. Some other presenters also referenced my presentation in their talks. Although, I couldn’t fully understand what they said (because it was in French). But, it made me feel good!

HEC and Quebec City tour

This was overall a relaxed week. I had decided to not do any more work and just focus on meeting new people and collecting ideas. Both of those goals were well achieved. I decided to enjoy the atmosphere. HEC Montreal (venue for the first conference) had quite a beautiful architecture. Both conference had breakfast included. I had croissant after a long time. Last time it was in 2019 when I visited Paris for a trip from US. I don’t cook such things at home.

The Quebec City seemed quite beautiful. It is a historical place. Heavy artillery guns surrounded the conference building. My friends helped me figure out what touristy places I could visit. I roamed around each day after the conference. I had a chance to visit the montmorency falls and the citadel. I decided to stay one more day at my own expense to do a helicopter tour. There is a flight school there that organizes “Pilot for a Day” experience. I decided to go for it. They taught me a few things about the flight and the controller. I had my first flying lesson, and we flew over the Quebec City. The citadel and the castle looked amazing!

Overall, I had a great experience and a lot of fun. I have collected a ton of new ideas and met many people in these conferences.

My Favorites:

Quote of the week: “Yet it is an iron rule of history that every imagined hierarchy disavows its fictional origins and claims to be natural and inevitable. For instance, many people who have viewed the hierarchy of free persons and slaves as natural and correct have argued that slavery is not a human invention.” — Yuval Noah Harari (from book Sapience)

Video of the week: Penrose Unilluminable Room