After 5 years of data collection, our WARN-D machine learning competition to forecast depression onset is now LIVE! We hope many of you will participate—we have incredibly rich data.
If you share a single thing of my lab this year, please make it this competition.
eiko-fried.com/warn-d-machi...
Captain Haddock looking very frazzled, saying, “What a week, huh?” Tintin leans into the frame and says, “Captain, it’s Wednesday.” Snowy aka Milou is very excited about the drink he has found.
How about starting the new year with a new Nilearn release?
Sure: here is Nilearn 0.13.0 for y'all!!!
New features: extract data to pandas or polars dataframes, get cluster statistics for surface analysis...
More info here: neurostars.org/t/nilearn-0-...
Would love to hear expert views on this paper. It appears to show that the operationalization of brain activity the field has relied on for 3 decades—the BOLD response—is not actually a sensible measure of brain activity.
www.nature.com/articles/s41...
Only use I have found so far in LLMs.
Helped me translate the recipes on my website.
remi-gau.github.io/recipe/en/oi...
Glad some people are willing to lose years of their life to listen, fact check and debunk this type of content. And also remain entertaining.
It actually increases maintenance burden, increases opportunity cost (I would have solved the problem faster), and decreases maintainers' willingness to engage with contributors.
You think AI is an awesome coding tool?
Have you tried reviewing a pull request that is being vibe-coded by someone not trying to engage with the problem at hand?
Episode 46: Sapiens
It's an ambitious goal to write the entire history of humanity in just 400 pages. It's even more ambitious to do it without reading any research.
Spent the afternoon cleaning my parents' garden, rather than cleaning someone else's mess in a codebase. #TouchGrass
Encountered way more bugs, but without ever having to get rid of them.
Never felt the urge to use dark mode. #vitaminD
Pruned some branches, but not the git kind.
Equally astonished. Same goes for monarchies that are still around. But then again I am French and we have historical grudges against monarchs and clerics.
The result of a large (42 authors!) collaboration:
"Go Figure: Transparency in neuroscience images preserves context and clarifies interpretation"
arxiv.org/abs/2504.07824
TL;DR: The FMRI world can (and should) improve results interpretation and reproducibility *today*, via transparent thresholding.
Only sometimes????!!!!
markets soar as investors realize that mommy didn't actually disappear, she was just hiding behind her hands
The problem with most machine-based random number generators is that they’re not TRULY random, so if you need genuine randomness it is sometimes necessary to link your code to an external random process like a physical noise source or the current rate of US tariffs on a given country.
Psych-DS is (1) spellcheck for your datasets and (2) a pathway to standardizing data in our academic fields that *everyone* can learn.
And it's live RIGHT NOW!
psych-ds.github.io
(This is the announcement post I've been leading up to)
If 80% is the threshold, I think people should get seizures from most websites with cooking recipes.
GitHub Action now officially support the free-threaded CPython.
hugovk.dev/blog/2025/fr...
If you care about computationally reproducible results but also like notebooks. You probably should ditch jupyter in favor of marimo.
Not if you include the cost of suffering my dad's jokes.
Today I was told I am more useful than a LLM. Take that, chatGPT!
WYOMING:
“Thank you, Madam chairman.”
“I prefer ‘Mister’ chairman.”
“Well you all voted preferred pronouns cannot be compelled speech.”
After years of research on human interaction and experimenting with novel ways to organize communities, today I unveil a brand new social media app.
The key idea: your interactions are arranged in ‘bins’ along a social dimension. You post, we put it in one of these ‘bins’.
The name? Histogram.
We checked how easy it is to match up people's brain maps after spatial normalisation. Turns out very! Maybe that surprising to many in the field - but my impression has long been that people are a bit in denial about deidentifying data. #neuroskyence #visionscience
doi.org/10.1101/2024...
Will check it out as I keep asking people: show me it can be done!
Though I have also heard people being stopped from sharing group level stat maps because they were told it could lead to individuals being reindentified. And that sounds wild to me.
Left my hand to my niece to do as she pleases and here is what happened.
Hope my fellow geeks will get all the video game reference.
Roy Baumeister called ego depletion "one of the most replicable findings in social psychology." As someone who spent 20 years studying it—and ultimately had to admit it wasn't real—I have to respectfully disagree. Here's my perspective of what went so wrong.
Friends, for something to be open source, we need to see
1. The data it was trained & evaluated on
2. The code
3. Model architecture
4. Model weights.
DeepSeek only gives 3, 4. And I'll see the day that anyone gives us #1 without being forced to do so, because all of them are stealing data.
I for one cannot wait for the next AI winter...
![Gojira - Mea Culpa (Ah! Ça ira!) [OFFICIAL VIDEO]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:7zdlopy7sm7ewbhlyfuhis64/bafkreiafx774yemt72xx7nvmxa32jwgzjvpcknfvyd74jcfsf7ejbpccza)
