Nicola Romanò @nicolaromano@qoto.org
- Pronouns
- he/him
Senior lecturer at Edinburgh University and Zhejiang-Edinburgh Joint Institute (ZJE).
Undergraduate Programme Coordinator, Biomedical Informatics at ZJE.
I teach #imageanalysis & #dataanalysis with #RStats & #python.
My research is focused on how #heterogeneous behaviour in #pituitary (and other) cells shapes their function as a population.
I'm also very interested in #reproducibility and #openscience.
Joined Oct 2022
Nicola Romanò
boosted
A Computational Approach to Interpreting the Embedding Space of Dimension Reduction https://www.biorxiv.org/content/10.1101/2024.06.23.600292v1?med=mas
I've recently picked up this great #book by Cathy O'Neil, "Weapons of Math Destruction"
https://books.google.co.uk/books?id=60n0DAAAQBAJ&newbks=0&printsec=frontcover&pg=PP1
An "old" book (2016) but still extremely if not more relevant than ever.
Nicola Romanò
boosted
Andrej Karpathy announced his upcoming LLM course - LLM101n, a few days ago. The course is not available yet, but the course syllabus is available on the course repo:
https://github.com/karpathy/LLM101n
There is no timeline for the release of this new course.
I've been reading again this little gem of a paper
I'm wondering how many results from #scrnaseq experiments are flawed for the reasons highlighted in there.
Nicola Romanò
boosted
Ok kids, here we go:
#Bandwagon, the open #Fediverse alternative to #Bandcamp is ready for its first steps into the light.
I have a waitlist online at
https://bandwagon.fm if you'd like to try it out.
I'm planning to build the first (dozen? I don't know) profiles in the waitlist personally, with an email questionnaire and FaceTime follow-ups if necessary.
Once I'm confident in the UX, we'll open up self-serve signups for everyone.
@RonBeavis @lgatto @petersuber Fine tuning is not so resource intensive as training, so journals could generate their own specific models without needing involvement of external companies. Not saying that would be a good idea, just that it's technically very feasible (also, there are plenty of open source models around)
Nicola Romanò
boosted
How many professional programmers are working on pointless and/or actively harmful products?
Give me your best guess.
(If you vote, please boost to diversify the results. It’s polite.)
Nicola Romanò
boosted
@scottwilson @b0rk I remember when people who wrote in Visual Basic were scoffed at but there was a lot of industry-specific software written by subject matter experts that wouldn't have been written at all without access to easy-to-learn and understand languages.
Nicola Romanò
boosted
If you are drafting figures for a scientific paper or presentation, remember that https://scidraw.io/ exists: a repository of free SVG cartoons for science.
#SciDraw is supported by the Sainsbury Wellcome Centre.
All content on SciDraw is shared under creative commons license (CC-BY) unless stated otherwise.
Nicola Romanò
boosted
Vaults are the most mysterious things inside the cell. This profile of Leonard Rome - the Vault Guy - is fascinating.
@drahardja actually it answers correctly, even with a biased prompt. Of course, it might have been fixed in the meantime or screenshot could be fake. Only way to tell would be to run a large number of queries with different photos and prompts.
@drahardja interestingly Google Images identifies the image correctly. I'm wondering what Gemini would have answered with an unbiased prompt...
Nicola Romanò
boosted
Check it out, the #Shiny extension for #VSCode now supports both Shiny for R apps *and* Shiny for Python apps in everyone's second favorite IDE!
Along with the updates come a few neat features for deploying your Shiny apps as serverless ShinyLive apps using https://shinylive.io, or for saving a ShinyLive app as local #RStats or #Python files.
Learn more and see the extension in action in our release post: https://shiny.posit.co/blog/posts/shiny-vscode-1.0.0/
Nicola Romanò
boosted
I am so deeply, deeply tired of every god damned "can be answered with a tutorial" technical question search being overwhelmingly flooded with videos. I don't want videos. I don't want to *listen* to something I can read. I am at least 50% deaf. GIVE ME WORDS DAMN YOU. WOOORDSSSSS!
@moorejh And an old favourite https://distill.pub/2016/misread-tsne/
@eamon I guess ffprobe should help https://ffmpeg.org/ffprobe.html
Nicola Romanò
boosted
I collected all my writing on PCA into a free long-form math book.
https://peterbloem.nl/publications/unraveling-pca
It's called "Unraveling principal component analysis" and it contains a narrative explanation of PCA all the way down to the foundations.
That means it discusses eigenvectors, singular value decompositions, the spectral theorem, low-rank decompositions and much more.
Nicola Romanò
boosted
The solar eclipse inspired our new puzzle design! It's double-sided: the sun on one side and the moon on the other. By flipping pieces over, you can gradually eclipse the sun turning day to night https://n-e-r-v-o-u-s.com/blog/?p=9636
Nicola Romanò
boosted
I put Breakout (aka Brick Breaker) inside Google Calendar! It lets you decline any meetings you shatter.
blog info + chrome extension link in thread!
@neuralreckoning @steveroyle I'm not advocating to replace thesis with publications, that would be problematic for a lot of reasons. I was purely talking about format. Given that the quality of the thesis is not really discriminant for getting a PhD, we might as well ask to do something that will be more useful for the student, but I'm sure that has downsides as well.
- Pronouns
- he/him
Senior lecturer at Edinburgh University and Zhejiang-Edinburgh Joint Institute (ZJE).
Undergraduate Programme Coordinator, Biomedical Informatics at ZJE.
I teach #imageanalysis & #dataanalysis with #RStats & #python.
My research is focused on how #heterogeneous behaviour in #pituitary (and other) cells shapes their function as a population.
I'm also very interested in #reproducibility and #openscience.
Joined Oct 2022
