NeuroData Lab
@neuro_data
Followers
8K
Following
5K
Media
115
Statuses
5K
We are devoted to discovering deep foundations linking all intelligences, in service to our collective flourishing (they/them). Joshua (jovo) Vogelstein.
here now
Joined March 2011
What is the 1 most impactful measurable/concrete goal that neuroscientists could collectively achieve in the next 5 years? . Please be as precise as possible, and do not include the word "understand".
14
11
27
RT @bridgen3rd: What are batch effects? How do they wreak havoc on our multi-site (neuroimaging) studies? 🤔 And where are current approac….
0
1
0
RT @Jere_je_je: 📢We all want to explain ML models, but what should these explanations convey, and how does one report these precisely and r….
0
1
0
Yay us! See everyone in a few days. .
Congratulations to all @JohnsHopkins researchers participating in #NeurIPS2024! Check out all @JohnsHopkins accepted papers, tutorials, and workshops at
0
0
1
Who is coming to @NeurIPSConf? DM me and let's meet up, I haven't been there in nearly a decade, excited to catch up with lots of people. .
0
0
1
AI Hype to Reality Translator. Learn —> Estimate.Hallucinate —> Error.LLM —> Next token predictor.Transformer —>Nadaraya–Watson kernel.Representation —> Summary statistics.Neural Network —> Nonlinear parametric function .Gradient descent —> Local linear approximation optimization.
1
2
8
RT @MilhamMichael: Could embracing analytical variability in neuroimaging be the key to more robust and generalizable scientific discoverie….
0
20
0
Come and work here, it is super fun!.
We’re thrilled to announce the #HopkinsDSAI Postdoctoral Fellowship Program! We’re looking for candidates across all areas of data science and AI, including science, health, medicine, the humanities, engineering, policy, and ethics. Apply today!.
0
1
0
RT @Jere_je_je: 🔥@JHUBME is hiring in Biomedical Data Science🔥Interested in problems in healthcare and data-driven solutions? Come join us….
0
8
0
"I was moderating hundreds of horrific and traumatising videos".. Anybody still using @OpenAI and thinks its morally ok?.
0
1
6
RT @alexisjbattle: Going on the faculty market this year? JHU is hiring at all levels. #ASHG2024 #ASHG24 folks in computational genomics,….
0
14
0
Israeli soccer fans attacked in Amsterdam, authorities say via @nbcnews . Not a huge surprise, given recent boycotting of Israeli's universities by Dutch universities. Israeli universities, by the way, are about 15% Muslim.
0
0
0
Remember it seems like it hasn't been updated in 5 years. What gives? Is there another site that is updated? Or as AI moved "beyond benchmarks" for some reason? Seems like maybe a bad idea, a la, @beenwrekt ?.
The dirty secret of machine learning is benchmarks are all we have.
0
0
1
RT @IrisVanRooij: I saw someone say a) they work on Ethical AI and b) use ChatGPT all the time. I’m baffled, given that ChatGPT is unethi….
0
144
0
And of course, our package, is available to reproduce all the figures, and to be used for hypothesis testing galore. Nearly all the work and credit is to Sambit Panda, We love questions and feedback of all kinds. 🧵/🧵.
0
0
4
Our new paper is out here: with the proofs and main simulations here: 6/🧵.
1
2
4
Of course, the procedures that work the best are those we previously published named "MGC" (short for multiscale-graph-correlation, and pronounced 'magic'). And of course, all the code is available and open source. 4/🧵.
2
0
4
Well, we wrote a paper which proves that there are actually many existing K-sample tests that are universally consistent, meaning they are statistical guaranteed to converge to the optimal, regardless of the dimensionality or distribution. 3/🧵.
1
1
3
If you are fancy, you could use a Hoteling test. But that also requires the distributions to be "nice" (eg, Gaussian). Also, it requires the sample size to be large relative to the number of dimensions (because it requires inverting an estimated covariance matrix). 2/🧵.
1
0
2