Ben Hardisty, PhD

Data mining on a scale you can't begin to fathom at Unknowm. Lover of 🥐☕️🐶 #useR #Rstats #maths #T1D @UofUmedicine

[Christopher W. Lynn @ChrisWLynn]

In short: Despite the apparent tension between simple descriptions and complex mechanisms, there is no contradiction. In fact, understanding how and why complex computations might appear simple is an exciting direction

There is still much to be done ✌️

[Christopher W. Lynn @ChrisWLynn]

2. Even if direct dependencies on synaptic inputs explain a large portion of the entropy, the remaining fraction (due in part to interactions between inputs) may still be critical for neural function

[Christopher W. Lynn @ChrisWLynn]

1. In large-scale recordings the synaptic inputs to each neuron are not known. I infer the optimal statistical inputs that carry the most info about each output, but these need not (and likely are not) the same

[Christopher W. Lynn @ChrisWLynn]

Does this mean that the mechanisms underlying a neuron's activity are simple? No. There are myriad ways that complex mechanisms might produce activity that is apparently simple (ie, explained by direct dependencies). To name two:

[Christopher W. Lynn @ChrisWLynn]

So what does this tell us about the brain? Precisely, it tells us that neural activity can be quantitatively predicted by simple models of artificial neurons, with only 10-40% of the entropy arising from high-order dependencies + latent stochasticity

[Christopher W. Lynn @ChrisWLynn]

The goal is to understand whether simple direct dependencies between neurons P(y|x_i) allow us to predict neural activity. Across several populations and species, these explain 60-90% of the entropy (variability) with no info about higher-order dependencies

[Christopher W. Lynn @ChrisWLynn]

There's been a lot of discourse surrounding my recent Nature Physics paper, to varying degrees of accuracy. Let me clarify the central conclusions and what I think they tell us about the brain
1/

https://t.co/vQtYVe94Ao

[Machine Learning (ML) Papers @Memoirs]

Midpoint Generative Models

Daniil Shlenskii, Nikita Gushchin, Lev Novitskiy, Dmitry V. Dylov, Alexander Korotin
https://t.co/iZi3QG33Rx [𝚌𝚜.𝙻𝙶]

[The Bulwark @BulwarkOnline]

Rep. Angie Craig: "Do you know how much money farmers lost last year?"

Brooke Rollins: *doesn't answer*

Craig: "It was $28 billion. You're the Secretary of Agriculture."

[Ben Hardisty, PhD @TomHardyofmath]

@hannahspierMD Oooh are there treatments for these now? I haven't heard about'em yet. Link to clinical trial data?

[Gigi @Gigi_chienergy]

The screwworm program wasn't charity, it was a $10 million fence that kept a billion-dollar problem from eating our own livestock alive. That's the thing with most USAID funding: it looks like "aid," but it's really cheap self-defense. Solve a problem there, and it never lands on our doorstep. But sure, let a bunch of guys who can't define DEI without Googling take a chainsaw to it. They didn't stop to ask, "Will this cut hurt us too?" Unless that's the point, burn it all down and call it efficiency.

As usual our inept agriculture secretary, blaming the Biden administration as usual, instead of offering a solution to the problem they created.

Texan here

A screwworm infestation is a nightmare for cattle, causing horrific wounds and economic devastation. For Texas, the situation has escalated dramatically in the last 24 hours with the first confirmed case of New World Screwworm (NWS) in over 60 years! This could have been avoided!

Infested animal can kill a cow in less than two weeks.

Treatment is extremely difficult and time-consuming, requiring the painful removal of every visible larva and deep disinfection of the wound. Ranchers no longer have much experience with this labor-intensive process, and there is currently no approved pharmaceutical treatment to make it easier.

This has triggered a massive economic threat. The USDA estimates that a widespread outbreak would drain an astonishing $1.8 billion from the Texas economy alone in livestock deaths, labor, and medication expenses.

How will this affect you? Tightening supplies will drive already high beef prices higher.

[DocJayBS-〽️D♿ @DocJayBSMD]

@ZainabSana2622 Criminal negligence causing death. Simple as that.

[Benthic Organism @BenthicOrganism]

@ZainabSana2622 Americans not even letting their kids live long enough to get shot at school

[🦢 @ZainabSana2622]

Father of Daisy, the second little girl to die of measles in Texas, tells Robert F. Kennedy Jr- founded anti-vaccine group, Children's Health Defense, that he doesn't regret not getting her vaccinated, and won't vaccinate any future children.

[Timothy Gowers @wtgowers @wtgowers]

It seems that LLMs could do the sum-product result too, even if humans in fact got there first. The details in this thread are pretty interesting.

[Laura Miers @LauraMiers]

I hate to be the bearer of bad news, but the recent Wired article was likely only the beginning of this summer propaganda wave.

Controversy & lies outperform truth every time. Struggling media outlets know exactly which buttons to push. There have been many such waves.

[Democrats Deliver @DemzDeliver]

🚨 Democrats have introduced a bill to eliminate all for-profit ICE detention centers.

[Stuart Buck @stuartbuck1]

@TECleveland @cremieuxrecueil I can see the DOGE tweet: "We canceled a contract for Panamanian screwworms, saving $165 million dollars!!!" [https://t.co/Bip6tKzkG1]

Meanwhile: "The slightest oversight could undo all the work that came before."

[Jack @tracewoodgrains]

screwworm prevention is one of those markers of state capacity that gives you confidence adults are in the room somewhere

one by one, those markers fall

[Zdenek Vrozina @ZdenekVrozina]

Almost one year after SARS2 infection, children with Long COVID showed measurable changes in the tiny blood vessels of the retina.
Wider arterioles.
Wider venules.
A shifted arteriole-to-venule ratio.
This was not just a symptom survey.
It was an objective microvascular signal🧵