Hi all! I am so excited to participate in this course again! Like many who’ve commented before me, I have benefited tremendously from the Fastai course and community since I first joined in 2018. It’s no stretch to say that my experiences here have been life-changing: I am now a tenure-track astronomer at the Space Telescope Science Institute and an associate research scientist at Johns Hopkins University. My research focuses on understanding how galaxies evolve by using a combination of astronomical observations and deep learning.
In previous threads I’ve said a bit more about my specific research projects, but maybe I’ll add another shameless plug… Earlier this year my collaborators and I used convolutional neural networks (a la Fastai) to distinguish rare, low-mass, nearby galaxies from the overwhelming number of faint, distant, background galaxies (paper here). The figure below shows our method in action: left panel shows examples from the training sample (~100k) and the right panel shows examples the test sample (>4M); upper panels show CNN-classified distant galaxies, while the lower panels shows nearby dwarf galaxies. This method allows us to study the relative distributions of “satellite” dwarf galaxies around more massive galaxies in a more comprehensive way than ever before!
Thanks again to Jeremy and the many others who have enabled a new (and growing) cohort of expert machine learning practioners! Feel free to get in touch with me on Twitter or LinkedIn if you’d like.