Time to start writing

During my long summer break I decided to start writing about my learnings on this website. Although I hope that my ramblings are helpful for others, I have to admit that my motivations to do so are selfish because:

  1. you truly master a new concept when you are able to explain it succinctly in your own words
  2. I have enjoyed reading over 85 papers in the last 2 years but lack an easy way to revisit my summary learnings from these efforts
  3. it's time to practise my English writing skills a bit

Now that you know why I am adding more online content let's spend a minute to talk about what I intend to write about. The short answer is “anything that I studied recently in my professional sphere of interest”. This starts from computer vision and may zoom in a lot on medical image analysis but I sometimes wander into methodology concepts that I fancy (e.g. Variational Auto-Encoders and causal inference to name two of my current favorites).

Next, let's talk briefly about my sources. Most of my study material are scientific papers published on arXiv/journals and papers or talks in conferences (e.g. MICCAI, ICML, NeurIPS, ICLR, …) or in readings groups that I regularly attend. You may ask how much I read and spend time studying new concepts, well much less than I would like but I try carving out a few hours per week in my schedule. I have tried following a ‘read two new papers every week’ routine for a while, but I have abandonned this by now. I currently follow my intuition browsing through the citation notifications that end up in my mailbox and diving into some when I am able to free up some time. More importantly, I decided to increase the depth of my readings and spend more time thinking about what I have read (favoring depth over breadth).

Finally, I would like to ask you, the reader, a favour. Please reach out to me for questions or comments on my content. I am specifically keen to find out whether my writings are clear and hear your views on or experiments with the posted content. Although this format is static and one-way, I would love to start a conversation with anyone interested!!

Avatar
Joeri Nicolaes
Computer Scientist / PhD student

My current research focuses on incorporating prior knowledge into deep neural networks and exploring how to make neural segmentation nets more shape-aware.