André Hechler

Mostly looking at brains @TU_muenchen. Currently working on neuroenergetics of cognition with

[Hadi Vafaii @hadivafaii]

How do brains “infer” the world’s state from noisy sensory data—and do so “dynamically?”

Our new theoretical framework bridges these two perspectives in a brain-inspired model👉🧵[1/n]

w/ amazing co-lead @dekelgalor & polymath mentor @jcbyts
📜preprint: https://t.co/xmjFLintZb

[artem kirsanov @ArtemKRSV]

Hopfield networks are a foundational idea in both neuroscience & ML 🧠💻
The latest video introduces the world of energy-based models and Hebbian learning 📹
A first step to covering Boltzmann/Helmholtz machines, Predictive coding & Free Energy later :)

https://t.co/sq8obkIpLy

[Luiz Pessoa @PessoaBrain]

"Functional connectivity" is simply a measure of correlation/covariance of signals. Does not measure communication between systems because we know that correlation can come about for lots of (too many) uninteresting reasons. Terrible name but stuck, other areas are using it too.

[André Hechler @andre_science]

@JamesMHarrison_ @PessoaBrain Our findings might also be in line with the whole entangled brain idea @PessoaBrain; the "energy saving" extends across both sensory and higher cognitive networks.

[André Hechler @andre_science]

@JamesMHarrison_ @PessoaBrain CMRO2 is the cerebral metabolic rate of oxygen consumption (in µmol/100g of GM tissue/min), which we measured with multiparametric quantitative BOLD, combing imaging of hemodynamics and oxygen extraction (2/2)

[Bobby Allyn @BobbyAllyn]

Statement from Scarlett Johansson on the OpenAI situation. Wow:

[Hamel Husain @HamelHusain]

Love this essay from @eugeneyan

This is especially acute for tools and infra around AI

[Luiz Pessoa @PessoaBrain]

𝗧𝗵𝗲 𝘁𝗶𝗽 𝗼𝗳 𝘁𝗵𝗲 𝗶𝗰𝗲𝗯𝗲𝗿𝗴: 𝗔 𝗰𝗮𝗹𝗹 𝘁𝗼 𝗲𝗺𝗯𝗿𝗮𝗰𝗲 𝗮𝗻𝘁𝗶-𝗹𝗼𝗰𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻𝗶𝘀𝗺 𝗶𝗻 𝗵𝘂𝗺𝗮𝗻 𝗻𝗲𝘂𝗿𝗼𝘀𝗰𝗶𝗲𝗻𝗰𝗲 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵
Happy to contribute a little to piece led by
@sNeuroble @psychonetrics @DScheinost
https://t.co/JXV8PPjDxI

[Grant Sanderson @3blue1brown]

The next chapter about transformers is up on YouTube, digging into the attention mechanism: https://t.co/TWNXiWM2az

The model works with vectors representing tokens (think words), and this is the mechanism that allows those vectors to take in meaning from context.

[Luiz Pessoa @PessoaBrain]

After 25 years of fMRI I no longer know how to analyze data... In more continuous paradigms at times all the brain "responds". ALL 85 threat-related regions got engaged when player runs away from chasing threat (t < 0) and enters safety (t > 0).
3 clusters of region responses:

[Erik Hoel @erikphoel]

Humans looking at LLMs

[François Chollet @fchollet]

Many people have heard the "in from three to eight years we'll have a machine with the general intelligence of a human being" quote from 1970, but much fewer have read the article it came from. It's a great read!

[André Hechler @andre_science]

@KordingLab @ProfData Once a model generates the correct results from a paraphrased description of background and methods (instead of an indirect classification based on weights), the timepoint might come closer.

[André Hechler @andre_science]

@ProfData @ken_lxl @ProfData: How likely do you think it is that the model learned the subtleties of human language when building up a certain type of result? Might a more powerful test be not to compare original vs. changed, but two paraphrased versions?

[bioRxiv Neuroscience @biorxiv_neursci]

Subjective confidence modulates individual BOLD patterns of predictive processing https://t.co/bcHvWzRcIy #biorxiv_neursci

[Teresa Anderson @1TeresaAnderson]

These might be the most terrifying graphs I have seen in my 20+years as a climate activist.

🌍

https://t.co/jYftstX87y

[bioRxiv Neuroscience @biorxiv_neursci]

The control costs of human brain dynamics https://t.co/Jo1VBPquDr #biorxiv_neursci

[bioRxiv Neuroscience @biorxiv_neursci]

Two distinct modes of hemodynamic responses in the human brain https://t.co/smCCv7hi5p #biorxiv_neursci

[bioRxiv Neuroscience @biorxiv_neursci]

The energy metabolic footprint of predictive processing in the human brain https://t.co/KyjQ0qT8rP #biorxiv_neursci