Corey Richier

Clinical psych PhD student at @UofIllinois. Psychopathology, comorbidity, machine learning, and systems/computational neuroscience. Opinions my own

[Kameron MacNear, M.S. @KA_MacNear]

So excited to continue my training with Michigan State University Counseling and Psychological Services! I couldn’t ask for a better fit between an internship program and my values as a clinician and supervisor.

Congratulations to all who matched today!

[Corey Richier @cjrichier]

@Carly_Lasagna Congrats!!

[Corey Richier @cjrichier]

Pleased to share another preprint! In this paper, we compare a suite of dimensionality reduction and feature selection methods for fMRI task decoding.

https://t.co/IbaZvyTU5q

[Corey Richier @cjrichier]

Happy to (finally!) be able to share a preprint of the first of papers from my dissertation! In this work we examined brain age prediction in individuals with and without GAD with structural MRI. https://t.co/8PWOkNRZS1

[Corey Richier @cjrichier]

@breannajrogers Congrats and welcome!!

[Corey Richier @cjrichier]

@paulcbogdan That is true, it’s nice that Conda can deal with CUDA. To me sometimes it feels like a hassle to have to make multiple environments just for a single project but I guess that’s not the end of the world 🤷🏼‍♂️

[Corey Richier @cjrichier]

I guess this is a sign to stop using conda?

[PyTorch @PyTorch]

We are announcing that PyTorch will stop publishing Anaconda packages on PyTorch’s official anaconda channels. For more information, please refer to the following post on dev-discuss: https://t.co/6gyNQ6Q8JP

[Ya'el Courtney, PhD @ScienceYael]

Just watched the full House hearing & testimony of NIH Director Dr. Monica Bertignolli.

A few takeaways:
• the proposed budget, in addition to not increasing for inflation, wants to condense NIH from 27 to 15 institutes. The NIH does not think this is a good idea.

🧵1/?

[Corey Richier @cjrichier]

To be fair my experience with conda has been that it always creates as many problems as it solves. Does anyone have any recommendations for package managers they like?

[Eric Simons @EricSimons]

My demons said I couldn’t do it,

To go from never running a marathon to full Ironman in <6mo, w/ a newborn & building/launching bolt․new,

But today after a 2.4mi swim, 112mi bike, full 26.2mi marathon, all back to back in 12h15m,

I am an Ironman.

[Miri Forbes @MiriForbes]

What happens when we break down DSM diagnoses into their individual symptoms and rebuild our classification system from the ground up? Do we see familiar DSM constructs, HiTOP dimensions, or something else entirely?

New paper out today! ✨ https://t.co/Om197YOdvl

1/🧵

[Corey Richier @cjrichier]

@kgates @ErinWestgate I agree! I think the tradition of favoring explanation in especially clinical areas holds back the implementation of potential interventions or clinical decision making processes. I wish there would be recognition that both elements are complimentary as opposed to competing

[Corey Richier @cjrichier]

@KSchaferPhD @kgates Supervised models are classifiers yes, but unsupervised learning models are not. The purpose of unsupervised learning is to uncover structure in data, which I’d argue is the same goal of statistical techniques such as factor analysis that are popular in psychology

[Yohan J. John @DrYohanJohn]

"Averaging over time hides what a neuron actually sees, what it actually gets to compute with."
💯
We need a handy review paper on artifacts of averaging. I come across wildly incorrect inferences from averaging & other data summaries way too often.
https://t.co/KvcQpfzGLR

[Hadas Weiss @weiss_hadas]

hands off! that’s the idea i planned to sit on for the next couple of decades

[JAMA Psychiatry @JAMAPsych]

This narrative review evaluates the validity of 9 diagnostic approaches in characterizing psychiatric symptoms.

https://t.co/WQ3BxQ1emO

[Mert R. Sabuncu 🤖🩻⚕️ @mertrory]

There is one thing I keep hearing from NIH study sections, in paper reviews, or as questions at seminars: deep learning is not intrepretable and this makes DL problematic for medical imaging. This is flawed and here's something to read about this: https://t.co/S0hgFmfyw9

[gavin leech (Non-Reasoning) @gleech]

New paper, listing 43 ways ML evaluations can be misleading or actively deceptive.

Following the good critics of psychological science we call these "questionable research practices" (QRPs). (The working title was "How To Lie In Machine Learning")