David DeMaris

research software engineer, spatio-temporal nested oscillations and dynamical neuroscience, postbop jazz piano

[David DeMaris @DavidDeMaris1]

@jim_rutt @brodeur The model I have (M3 16", 36g) happily runs all the local LLMs I've thrown at it, deepseek r0 and gemma most recently. I DJ'd the EFF party a few weeks ago with fancy "unmixing AI" software while talking to deepseek about designing a new cryptography algorithm, no glitches.

[David DeMaris @DavidDeMaris1]

@MillerLabMIT For clarity, it 's just another signal processing landmark in the historical review. But it's the rare work on IT using systematic stimuli and proposing a frequency code rather than grandmother cell or high level feature specific cell.

[David DeMaris @DavidDeMaris1]

@MillerLabMIT I'd like to have seen a cite of the late 80s/90s Richmond and Optican papers on Temporal Encoding of Two Dimensional Patterns (multiplex filter hypothesis). Aperiodic activity in IT - they were influential for me doing nonliear wave interaction image processing.

[David DeMaris @DavidDeMaris1]

@t_andy_keller @mozesjacobs You might be interested in work I did in engineering at UT Austin 1995-2001 which bears some similarities, but in a nonlinear dynamics framework. Pretty sure I cited that sound of a drum paper.
https://t.co/gH5pjdzZHz

[David DeMaris @DavidDeMaris1]

@MillerLabMIT Link seems broken?

[bioRxiv Neuroscience @biorxiv_neursci]

Abstract, modality-specific and experience-dependent coding of affect in the human brain https://t.co/qQaCiwEPn2 #biorxiv_neursci

[bioRxiv Neuroscience @biorxiv_neursci]

Non-Stationary Dynamic Mode Decomposition https://t.co/zeIBkovUxG #biorxiv_neursci

[David DeMaris @DavidDeMaris1]

Always like to see *computing* with waves in RNN

[Andy Keller @t_andy_keller]

Traveling waves are known to exist throughout the brain in a variety of forms — there are many hypotheses, but their exact computational role is debated.

Together with @wellingmax we built an RNN which exhibits traveling waves to see what it could do. Here’s what we think:

1/7

[David DeMaris @DavidDeMaris1]

X where the bird used to be but still says tweet. Shouldn't it be xeet now?

[Nima Dehghani @neurovium]

Follow up on promises! 📢
"Symmetry making and symmetry breaking in cortex A collective portrait of ensemble excitation and inhibition"
https://t.co/BuKMsVG1lD
(this is paper 1 from "symmetry breaking/spatiotemporal series" series)
see 🧵👇1/n

[Shaoshi Zhang @ZShaoshi]

1/11 Super excited to share our latest preprint! 📢

We estimate a putative marker of whole-cortex excitation/inhibition (E/I) ratio that indexes cortical maturation in youth! Leveraging a large-scale biophysically plausible model, we can estimate E/I ratio in-vivo! 🧵

[David DeMaris @DavidDeMaris1]

More than once I've finger wagged at youngsters failing to do their research on precedents. I just found some work from '86 that I ought to have cited.
https://t.co/wcTxYVlqlT

[David DeMaris @DavidDeMaris1]

@neurograce There's an earlier paper "The Siren Call of Metaphor" circa 2004.

[David DeMaris @DavidDeMaris1]

@neurograce The late Gerhard Werner, O.G. single unit physiologist with Mountcastle, wrote a few critical reviews. I think this was the last.
https://t.co/SzijSTTw6r

[David DeMaris @DavidDeMaris1]

@sungho_hong_ @TonyZador Perhaps a little like symbolic AI failing on the endless edge cases?

[David DeMaris @DavidDeMaris1]

@lathropa "Rewrite this function, but with hard to detect bugs"

[David DeMaris @DavidDeMaris1]

@realpotofgreed Glad I read that preprint a few weeks ago by Bongard and Levin on polycomputing, so that I can feel good about saying all of the above on the same hardware.
https://t.co/0XhjmQ6ekg

[David DeMaris @DavidDeMaris1]

@IntuitMachine @OpenAI Possibly relevant, see esp. the part about the institutional review committee. DH says there's never been any conflict between their take and the parent.
https://t.co/5uT61C7AxU

[David DeMaris @DavidDeMaris1]

@pickover I can roll the paper into a cylinder two ways so I'm thinking 7 choose 3 plus 5 choose 3 plus 6 choose three.

[David DeMaris @DavidDeMaris1]

For me it was natural to think about low frequencies modulating neural field parameters like nonlinearity (bifurcation) parameters, coupling constants in laterally coupled neural fields. Mostly I see people talk about inhibition, gain modulation - circuit ideas.

[David DeMaris @DavidDeMaris1]

Any other people in theoretical or computational neuroscience come in with a background in sythesizers, especially modular synthesis? Did it affect the way you think about things?
Just curious.