mer. 05 janvier 2022

Hi all, and happy New Year!

2021 has ended and what a better time to debrief everything that went through this semi-awful year. I've tried to keep up an account of the achievements of the year, although I started in the spring so a few things might have slipped through. Oh, well, this is fine.

Publish or perish: still alive!

I have worked on five papers this year. Three have been already published:

• PKSpell: Data-driven pitch spelling and key signature estimation, in ISMIR 2021.
• Web Image Context Extraction with Graph Neural Networks and Sentence Embeddings on the DOM tree, in the GEM workshop of ECML/PKDD 2021.
• Robust and Decomposable Average Precision for Image Retrieval, in NeurIPS 2021.

A fourth one is going to be published in the beginning of next year in WCNC.

Overall, a nice track record. I dabbled in a lot of different things in 2020/2021 and this shows in the venues I've submitted to: musical information retrieval, graph machine learning, wireless communications… However I feel that my contribution to most of those papers was quite limited: the real work was done by graduate students or postdocs who deserve all the credit! I'd like to be a little bit more focused this year so that I can really put my hands-on some research, maybe with a solo publication. I've got a few ideas but it's hard to find time to dive into it properly.

Note that I've got two papers rejected this year. One is going to be resubmitted, the other was dropped as the research was still too fresh. This is part of the game but it still stings a little bit when it happens to you!

Project and grants: still a small fish

I am pleased to work on two projects that have a small funding (you know, enough to hire an intern):

• Reinforcement learning for games, with colleagues from the ILJ team.
• Super-resolution of remote sensing time series, with Charlotte Pelletier.

These are small scale projects… however, they are mine. :-) Which is very gratifying. Both are still ongoing in 2022 so I've got this to look forward to.

I've also submitted a preproposal for a "young researcher grant" from ANR, the French funding agency for research. I'm not optimistic on the outcome but you never know.

Students, interns and so on

My two PhD students were in their first year so they still have a lot of way in front of them! It's great to see them evolve and grow as researchers though. They wrote their first paper and faced the review process for the first time. Keep going!

I was the main advisor for two interns this year, which gobbled up a lot of time. Covid and remote working did not make things easier and I am still fumbling for a effective way to manage students using Slack and Teams meetings only. Anyway, the internships went well and I am pleased of what we accomplished.

Awards, distinctions, bonuses: still not excellent enough

Although I've gotten some small-ish runner-ups or accessits to awards before, I'm not used to receive distinctions. This year was no surprise! I applied to get the bonus for "excellent research" from the French research ministry and was denied, mostly because none of my PhD students have defended yet.

On the bright side, I was distinguished as an outstanding reviewer for ICCV and BMVC this year. I review around 30 papers every year and, while it is somewhat of a chore, I try my best to deliver useful feedback to authors. So I'm glad that people trust me as reviewer. :-)

Teaching: still fun… except Covid

We have created new courses this year and I was responsible for the modules on deep generative models and deep learning for audio. This was a blast to create and teach these classes. Yet… this was also a huge pain. I am starting to develop the feeling that I am skilled in designing practical exercices but not a great lecturer. Due to Covid, most classes were pulled back and we had to teach remotely, which is a terribly poor way to engage students.

Overall, teaching in 2021 was disappointing, at best. However, we've gone back to teaching on-premises since september so things are looking up.

Extra-stuff: still a mixed bag

I've dropped the C++ class I taught at the École nationale des ponts et chaussées. It pains me since I loved this class but I simply did not have the time to do it anymore and it was becoming tiring due to calendar mismatches¹.

Apart from the usual reviews for papers and journals, I also did reviews of grant proposals for various funding agencies. This is definitely not my favorite activity, but some of these reviews are interesting and foreign funding agencies generally offer a compensation of a few hundred euros for the time spent. Not always worth the time but not a complete waste either.

Conclusion

I am still adjusting to my role as an associate professor. Balancing teaching and research is hard. Painfully hard, sometimes. Despite my taste for teaching, I somewhat would like to reduce my number of classes so I have the time to be more hands-on with my research. Being a advisor is enjoyable but I miss scratching my own itches! Job-wise, I'm also pessimistic in face of the many reforms that the government is putting in place. Call me vain but I think I deserve a higher salary and a better job environment. Fighting for scraps of grants and the death by a thousand cuts that is the administration are starting to tire me down. And I'm only two and half years in! Anyway.

Covid sucks. Remote work is… not for me. I can deal with a one or two remote days a week. More? No way. Working from home is depressing as a researcher: I like to discuss with colleagues, to bounce off ideas or just chat about what's going on. Maybe it would be fine if I was working alone, I'm just not keen on being a remote only advisor. Teaching remotely however is starting to turn into my own personal hell, especially for the practical classes that are the bulk of my schedule. I hate it. Damn pandemic.

2021 was not an amazing year. It was fine on the professional side, i.e. we made do. I am hopeful that things will clear up a little bit so that I can be more focused this year. Maybe Covid will start receding at some point too. We can dream.

Best wishes!

Category: meta