1. Keep my research under a bigger picture
It might easily loose the sense of what I am doing and why I am doing what I am doing. One tip that works for me to prevent that feeling is to put the research into a bigger picture, so even when things don't work out, I still feel that I am still contributing towards my goals. For me, I am interested in optimal learning and reinforcement learning, especially under adversarial contexts. It also might help to write an editable research statement and regularly blog on the topics.
2. Keep myself updated on the cutting edge of research in Machine Learning and Artificial Intelligence (MLAI)
One concrete activity for that is I like to summarize ideas from recently accepted papers in top-tier conferences (currently I pick ICML, NIPS, and AISTATS)
3. Keep a notebook of quick ideas
When reading new papers and attending presentations, I write down good ideas that are potentially helpful for my own problems.
4. Switching between research and "other works" when I feel low
A good example of "other works" for me is to do 2 or check Twitter and/or Reddit to update about research.
Update #1: This is really challenging to keep a good pave because there are many papers and I do not have much free time for all this, but still I want to see how much of it I can implement. Maybe it is hard to write summary on all published papers, I can instead focus on some while take a quick look on the others. In addition, one practical tip I learn today is that, instead of use Google Sheet to keep of ideas of recently published papers in top-tier conferences, I instead will, adapting the idea of David Abel (https://david-abel.github.io/notes.html), write in a latex format as an informal report. The main difference is that David Abel attends the conferences and seems to write out of the representations from the authors, while I am writing directly from their papers.
ReplyDelete